University of Southern California Dissertations and Theses

Quantify human experience: integrating virtual reality, biometric sensors, and machine learning

(USC Thesis Other)

Quantify human experience: integrating virtual reality, biometric sensors, and machine learning

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

QUANTIFY HUMAN EXPERIENCE:
INTEGRATING VIRTUAL REALITY, BIOMETRIC SENSORS,
AND MACHINE LEARNING

by

Dejian Chen

A Thesis Presented to the
FACULTY OF THE USC SCHOOL OF ARCHITECTURE
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF BUILDING SCIENCE

May 2022

Copyright 2022 Dejian Chen
i

Acknowledgements

I would like to express my great gratitude to my parents, who have been supported my study
at USC.

I would like to thank my thesis chair Professor Joon-Ho Choi for his guidance and supports.

I would like to express my appreciation to my thesis committee member Professor Karen
Kensek. I could not have gone this far without her expertise, patient guidance, and valuable advice.

I want to thank my thesis committee member Professor Yan Liu who gave me valuable
advice.

I want to specially thank Professor Douglas Noble for his guidance and encouragement at
the beginning of the work.

I want to thank all participants in my study. I could not have completed this study without
your helps.

ii

Joon-Ho Choi, Ph.D. LEED AP BD+C
Associate Dean of Research & Creative Work; Associate Professor, School of Architecture,
University of Southern California.

Karen Kensek, LEED AP BD+C
Professor of Practice. ACSA Distinguished Professor, School of Architecture, University of
Southern California.

Yan Liu, Ph.D.
Associate Professor, Viterbi School of Engineering, University of Southern California.

iii

TABLE OF CONTENTS
Acknowledgements ............................................................................................................................................. i
List of Tables ...................................................................................................................................................... v
List of Figures ................................................................................................................................................... vii
Abstract ............................................................................................................................................................... x
Chapter 1 Introduction ..................................................................................................................................... 1
1.1 Office indoor environment and occupant productivity ................................................................. 2
1.1.1 IEQ and Building Energy Cost ............................................................................................... 2
1.1.2 IEQ and Financial Benefits ..................................................................................................... 3
1.1.3 IEQ and Human Well-being ................................................................................................... 4
1.1.4 IEQ and occupants’ productivity ........................................................................................... 5
1.1.5 Factors that affect comfort and productivity ........................................................................ 6
1.2 Neuroscience and Architecture .......................................................................................................... 7
1.2.1 Neuroscience as an emerging tool to quantify human experience .................................... 7
1.2.2 Neuroscience and Autonomic Nervous System .................................................................. 8
1.2.3 Quantify human experience through biometric sensors ..................................................... 9
1.3 Virtual Reality...................................................................................................................................... 10
1.3.1 What is Virtual Reality? .......................................................................................................... 10
1.3.2 A brief history of Virtual Reality .......................................................................................... 13
1.3.3 The use of Virtual Reality in psychology study .................................................................. 14
1.3.4 Bring Virtual Reality into the Architecture ......................................................................... 15
1.4 Machine Learning ............................................................................................................................... 16
1.4.1 What is Machine Learning? ................................................................................................... 17
1.4.2 Supervised Learning Method ................................................................................................ 19
1.5 Summary .............................................................................................................................................. 20
Chapter 2 Background and literature review ...............................................................................................21
2.1 Design Features that Affect Occupant Well-being and Productivity ......................................... 21
2.1.1 Daylight and Lighting ............................................................................................................. 22
2.1.2 Biophilia and views ................................................................................................................. 24
2.1.3 Look and feel ........................................................................................................................... 25
2.2 Physiological background .................................................................................................................. 26
2.2.1 How do humans sense emotions? ........................................................................................ 26
2.2.2 Measure ANS........................................................................................................................... 27
2.2.3 Heart rate and heart rate variability ...................................................................................... 28
2.2.4 Electrodermal activity and skin temperature ...................................................................... 29
2.3 Virtual Reality...................................................................................................................................... 31
2.3.1 Ecological Validity of VR ...................................................................................................... 32
2.3.2 Presence, Immersion, and Measure of Presence ................................................................ 33
2.3.3 Studies of presence in the architectural domain ................................................................. 34
2.4 Summary .............................................................................................................................................. 36
Chapter 3 Methodology ..................................................................................................................................38
iv

3.1 Experiment Preparation .................................................................................................................... 39
3.1.1 Identify Key Architectural Design Features and Their Variables .................................... 40
3.1.2 Rendering workflow of office model in Unity ................................................................... 40
3.1.3 Generate VR office scenes .................................................................................................... 48
3.1.4 Measure of Physiological Responses and Subjective Assessment ................................... 54
3.1.5 Data Synchronization ............................................................................................................. 57
3.2 Experiment Procedure....................................................................................................................... 58
3.3 The COVID-19 Guidelines .............................................................................................................. 63
3.4 Data analysis ........................................................................................................................................ 64
3.5 Summary .............................................................................................................................................. 72
Chapter 4 Data sample and data synchronization .......................................................................................74
4.1 Data Sample and Raw Data of Questionnaires ............................................................................. 74
4.1.1 Demographic questionnaire .................................................................................................. 74
4.1.2 In-VR Questionnaire of Office Assessments ..................................................................... 76
4.1.3 Questionnaire of Presence ..................................................................................................... 78
4.2 Data Sample and Raw Data of Biometric data .............................................................................. 79
4.2.1 Heart Rate and Stress Level................................................................................................... 79
4.2.2 Electrodermal activity ............................................................................................................. 82
4.2.3 Skin Temperature .................................................................................................................... 87
4.3 Data for machine learning ................................................................................................................. 88
4.4 Summary .............................................................................................................................................. 91
Chapter 5 Results and discussion ..................................................................................................................93
5.1 Analysis of the in-VR questionnaire ................................................................................................ 93
5.2 Analysis of biometric data .............................................................................................................. 103
5.3 Predicting participants’ preferences using machine learning models ...................................... 133
5.4 Analysis of the questionnaire of presence ................................................................................... 137
5.5 Summary ........................................................................................................................................... 140
Chapter 6 Conclusion and future work ..................................................................................................... 143
6.1 Conclusion ........................................................................................................................................ 143
6.2 Research limitation and future work ............................................................................................ 146
6.2.1 Better rendering and VR experience ................................................................................. 146
6.2.2 Size, quality, and dimensions of the data .......................................................................... 147
6.2.3 Diverse study objectives ..................................................................................................... 147
6.2.4 A more controlled lab setting ............................................................................................. 148
6.2.5 Comparison of real versus virtual environments ............................................................ 148
6.3 Summary ........................................................................................................................................... 149
Bibliography ................................................................................................................................................... 150
Appendices ..................................................................................................................................................... 160

v

LIST OF TABLES
Table 2-1 Literature review of design features and their impact on occupants ...................................... 21
Table 3-1 Design features and variables ....................................................................................................... 40
Table 3-2 Oculus Quest 2 Hardware Specifications (Oculus 2020) ......................................................... 49
Table 3-3 Demographic survey ...................................................................................................................... 56
Table 3-4 Questionnaire: Participants’ evaluation of each feature ........................................................... 56
Table 3-5 Final questionnaire ......................................................................................................................... 57
Table 3-6 Event-related Analysis Variables (Benedek and Kaernbach 2010) ......................................... 67
Table 4-1 Demographic questionnaire ......................................................................................................... 75
Table 4-2 Data sample of the in-VR questionnaire .................................................................................... 77
Table 4-3 Data sample of questionnaire of presence ................................................................................. 78
Table 4-4 Original heartrate and stress level of subject 06 ........................................................................ 80
Table 4-5 Event time of subject 06 ............................................................................................................... 80
Table 4-6 Rounded heartrate for each event (subject 06) .......................................................................... 81
Table 4-7 raw data and data synchronization of subject 06 ....................................................................... 82
Table 4-8 EDA data sample and data preprocess of subject 06 ............................................................... 83
Table 4-9 EDA in the format for Ledalab (subject 06) .............................................................................. 85
Table 4-10 Data sample of skin temperature ............................................................................................... 87
Table 4-11 Mean skin temperature of each event (subject 06).................................................................. 88
Table 4-12 Top 5 rows of the integrated data ............................................................................................. 91
Table 5-1 Wilcoxon signed-rank test of lighting source ............................................................................. 95
Table 5-2 Wilcoxon signed-rank test of Illuminance level......................................................................... 97
Table 5-3 Wilcoxon signed-rank test of window view ............................................................................... 99
Table 5-4 Wilcoxon signed-rank test of indoor plants ............................................................................... 99
Table 5-5 Wilcoxon signed-rank test of color .......................................................................................... 101
Table 5-6 Wilcoxon signed-rank test of texture ....................................................................................... 102
Table 5-7 Wilcoxon signed-rank test of ceiling height ............................................................................ 103
Table 5-8 paired t-test of lighting source ................................................................................................... 107
Table 5-9 sign test of lighting source ........................................................................................................ 107
Table 5-10 paired t-test of illuminance level ............................................................................................. 112
Table 5-11 sign test of illuminance level.................................................................................................... 112
Table 5-12 Paired t-test of view .................................................................................................................. 116
Table 5-13 sign test of view ........................................................................................................................ 116
Table 5-14 Paired t-test of plants ............................................................................................................... 120
Table 5-15 sign test of plants ...................................................................................................................... 120
Table 5-16 paired t-test of color ................................................................................................................. 124
Table 5-17 sign test of color ........................................................................................................................ 124
Table 5-18 paired t-test of texture .............................................................................................................. 125
Table 5-19 sign test of texture ..................................................................................................................... 125
Table 5-20 paired t-test of ceiling height ................................................................................................... 133
Table 5-21 sign test of ceiling height ......................................................................................................... 133
Table 5-22 The best parameters of logistic regression ............................................................................ 134
Table 5-23 The best parameters of SVC ................................................................................................... 134
Table 5-24 The best parameters of random forest .................................................................................. 135
Table 5-25 The best parameters of articial neural network .................................................................... 135
Table 5-26 The classification report in terms of weighted score ........................................................... 137
vi

Table 6-1 Summary of subjective evaluation and biometric data .......................................................... 145

vii

LIST OF FIGURES
Figure 1-1 Roles of the sympathetic and parasympathetic systems. Redrawn from (Jänig 1989). ........ 9
Figure 1-2 reality virtuality continuum. Redrawn from (Milgram et al. 1995). ....................................... 11
Figure 1-3 VR Head-mounted Display, Dec 31, 2019, Insider, (Insider 2019) ...................................... 12
Figure 1-4 Semi-immersive VR, n.d., visbox, (Visbox n.d.) ...................................................................... 12
Figure 1-5 VR on smartphone, 2017, (Bnext3D 2017) .............................................................................. 13
Figure 1-6 VR in architecture, June 20, 2020, ARK Architects, (ARK architecture 2020) ................... 16
Figure 1-7 Machine learning technologies, Jan 11, 2022, EURIX, (EURIX 2022) ............................... 18
Figure 1-8 machine learning for smart building applications (Djenouri et al. 2019) ............................. 19
Figure 2-1 Emotion elicitation model, (Levenson 2014) ........................................................................... 27
Figure 2-2 Two dimensions of valence and arousal. (Wang, Nie, and Lu 2014) .................................... 28
Figure 2-3 EDA is composed of Tonic EDA and Phasic SCRs (redrawn from (Benedek and
Kaernbach 2010)) ............................................................................................................................................. 30
Figure 2-4 Skin Conductance data consists of phasic SC (steep change) and Tonic SC (slow
change) (analysis of subject 07 in Ledalab) .................................................................................................. 30
Figure 2-5 Virtual building evacuation study (Zou, Li, and Cao 2017).................................................... 34
Figure 2-6 Exploring user experience RE (left) VE (right) (Kuliga et al. 2015) ..................................... 36
Figure 2-7 VR for perception of daylit RE (left) VE (right) (Kynthia Chamilothori, Wienold,
and Andersen 2019) ......................................................................................................................................... 36
Figure 3-1 Methodology Diagram ................................................................................................................. 39
Figure 3-2 Floor plan of the office in Revit ................................................................................................. 41
Figure 3-3 Area light and light settings ......................................................................................................... 42
Figure 3-4 Based lightmap of one room ....................................................................................................... 42
Figure 3-5 Global Volume settings ............................................................................................................... 44
Figure 3-6 HDRI of landscape view, n.d., PolyHaven, (Majboroda n.d.) ............................................... 44
Figure 3-7 Material setting of glass ................................................................................................................ 45
Figure 3-8 emissive material and bloom, no bloom effect (left), bloom effect (right). ......................... 46
Figure 3-9 lightmap setting ............................................................................................................................. 46
Figure 3-10 Baked reflection probe cast reflection on reflective surface; no reflection probe
(upper one) ........................................................................................................................................................ 47
Figure 3-11 Office Model in Unity ................................................................................................................ 47
Figure 3-12 Workflow of generating VR scenes ......................................................................................... 49
Figure 3-13 XR rig and UI interactor on the left-hand controller ............................................................ 50
Figure 3-14 Teleport move ............................................................................................................................. 50
Figure 3-15 Subscribe and unsubscribe method ActivateUIMode........................................................... 51
Figure 3-16 Colliders of furniture in the scene ............................................................................................ 52
Figure 3-17 Participant uses hand to push a door ...................................................................................... 52
Figure 3-18 How the enter time is logged. (The cylinder represents the user. The console logs
the time when the user entered Trigger Area.) ............................................................................................ 53
Figure 3-19 Specification of Biometric Sensors .......................................................................................... 55
Figure 3-20 Example of a JSON object ....................................................................................................... 57
Figure 3-21 The investigator tests the program in the lab (wearing mask and face cover) .................. 58
Figure 3-22 Design features and variables of 14 offices ............................................................................ 61
Figure 3-23 UI instruction and feature reminder ........................................................................................ 62
Figure 3-24 Questionnaire in VR .................................................................................................................. 62
Figure 3-25 Experiment Procedure. .............................................................................................................. 63
viii

Figure 3-26 Disposable VR face cover ......................................................................................................... 64
Figure 3-27 EDA analysis in Ledalab ........................................................................................................... 66
Figure 3-28 Event-related plot of subject 06 (time window -1 to 5 seconds; features: plants
and ceiling height) ............................................................................................................................................ 68
Figure 3-29 machine learning workflow and future works ........................................................................ 70
Figure 3-30 Trade-off between bias and variance. (VanderPlas 2016) .................................................... 71
Figure 3-31 K-fold cross validation (scikitlearn.org) .................................................................................. 71
Figure 3-32 confusion matrix (Müller, A., & Guido 2018) ........................................................................ 72
Figure 4-1 The heatmap of Pearson correlation of numerical values ...................................................... 90
Figure 5-1 Interval plot of lighting source ................................................................................................... 94
Figure 5-2 Interval plot of illuminance level ................................................................................................ 96
Figure 5-3 Interval plot of window view ...................................................................................................... 97
Figure 5-4 Interval plot of indoor plants ...................................................................................................... 98
Figure 5-5 Interval plot of color ................................................................................................................. 100
Figure 5-6 Interval plot of texture .............................................................................................................. 101
Figure 5-7 Interval plot of ceiling height ................................................................................................... 102
Figure 5-8 Event-related responses of lighting source ............................................................................ 104
Figure 5-9 mean value of ISCR (lighting source) ..................................................................................... 105
Figure 5-10 mean value of heart rate (lighting source) ........................................................................... 105
Figure 5-11 mean value of stress level (lighting source) ........................................................................ 106
Figure 5-12 mean value of skin temperature (lighting source) .............................................................. 106
Figure 5-13 Event-related responses of illuminance level ...................................................................... 109
Figure 5-14 mean value of ISCR (illuminance level) ............................................................................... 110
Figure 5-15 mean value of heart rate (illuminance level) ........................................................................ 110
Figure 5-16 mean value of skin temperature (illuminance level) ........................................................... 111
Figure 5-17 mean value of stress level (illuminance level) ...................................................................... 111
Figure 5-18 Event-related responses of view ........................................................................................... 113
Figure 5-19 mean value of ISCR (view) ..................................................................................................... 114
Figure 5-20 mean value of heart rate (view) ............................................................................................. 114
Figure 5-21 mean value of stress level (view) ........................................................................................... 115
Figure 5-22 mean value of skin temperature (view) ................................................................................. 115
Figure 5-23 Event-related responses of indoor plants ............................................................................ 117
Figure 5-24 mean value of ISCR (plants) .................................................................................................. 118
Figure 5-25 mean value of heart rate (plants) ........................................................................................... 118
Figure 5-26 mean value of stress level (plants) ......................................................................................... 119
Figure 5-27 mean value of skin temperature (plants) .............................................................................. 119
Figure 5-28 Event-related responses of color........................................................................................... 121
Figure 5-29 mean value of ISCR (color) .................................................................................................... 122
Figure 5-30 mean value of heart rate (color) ............................................................................................ 122
Figure 5-31 mean value of stress level (color) .......................................................................................... 123
Figure 5-32 mean value of stress level (color) .......................................................................................... 123
Figure 5-33 Event-related responses of texture ....................................................................................... 126
Figure 5-34 mean value of ISCR (texture) ................................................................................................ 127
Figure 5-35 mean value of heart rate (texture) ......................................................................................... 127
Figure 5-36 mean value of stress level (texture) ....................................................................................... 128
Figure 5-37 mean value of skin temperature (texture) ............................................................................ 128
Figure 5-38 Event-related responses of ceiling height ............................................................................ 130
ix

Figure 5-39 mean value of ISCR (ceiling height) .................................................................................... 131
Figure 5-40 mean value of heart rate (ceiling height) ............................................................................. 131
Figure 5-41 mean value of stress level (ceiling height) ........................................................................... 132
Figure 5-42 mean value of skin temperature (ceiling height) ................................................................ 132
Figure 5-43 The converging process of the artificial neural network ................................................... 135
Figure 5-44 ROC curve of models ............................................................................................................. 137
Figure 5-45 The questionnaire of presence.q1: “to what extent did you have a sense of being
in the office? (Not at all -2, -1, 0, 1, 2 very much so); q2: how much time did you feel the
office is 'real,' and you forgot the 'real world'? (Never -2, -1, 0, 1, 2 almost all the time); q3:
when you think back on your experience, do you think of the office as images that you saw or
more as somewhere that you visit? (Only as images -2, -1, 0, 1, 2 somewhere I visited).” ................ 138
Figure 5-46 participants' feedbacks on improving the sense of presence ............................................ 139
Figure 5-47 count of feedbacks from participants ................................................................................... 140

x

Abstract
A new approach that integrates virtual reality and biosensors was proposed to investigate the
impacts of interior architectural design features on occupants’ subjective perceptions and
physiological responses. Seven design features (lighting source, illuminance level, window view,
indoor plants, color, texture, and ceiling height) were selected through literature reviews. A virtual
environment configured by fourteen offices was created by Revit and Unity. Each pair of offices
contain two different design variables from the same design feature. 32 participants joined the study
and navigated through 14 offices using VR headset. Subjective evaluations (satisfaction, stress level,
motivation to work, concentration level) and biometric data including heart rate, stress level (heart
rate variability), electrodermal activities, and skin temperature of participants were collected during
the experiment.
The findings show participants’ preferences over daylight, green window view, indoor plants,
bright and beige color. Statistically significant differences in skin conductance response were
observed between daylight and artificial lighting. Marginally significant differences in heart rate were
found between plants and no plants. Machine learning was adopted to test the applicability of the
datasets. The results reveal that the random forest model performs best at predicting participants’
preferences (accuracy: 0.73). The analysis of the questionnaire of presence indicated sufficient
ecological validity of the study. The proposed method provides a systematic way for designers to
evaluate occupants’ preferences of design features. It demonstrated the potential to quantify the
human experience of design features in terms of biometric data.

xi

Hypothesis:
Subjects exhibit the same preference over a specific variable in a virtual office environment
among the architectural design features concluded from the literature review.
The second part investigates if the differences of biometric data between two design
variables of each feature are statistically significant.
The users’ preferences can be predicted by machine learning models using biometric data.

Research objectives:
To develop a data analytics method to understand the relationship between the users’
physiological response and design variables in virtual environment.
To propose a preference model as a function of design features and biometric data.
To develop a novel preference model that validates the feasibility of using biometric data to
quantify the human experience.

Key words:
Virtual reality, biometric data, indoor environment, machine learning, architectural design,
biosensor

1

Chapter 1 Introduction
It is widely accepted that indoor environment quality (IEQ) is closely associated with
building energy use, financial profits, occupants’ health, well-being, and productivity (W. J. Fisk
1997; W. Fisk 2002; Al Horr et al. 2016). Architectural design features as one of the main
components of an indoor environment are features that configure an indoor space and can be
visually identified. Understanding the relationships between human experience and architectural
design features is critical because the knowledge will help designers shape an indoor environment
that reduces occupants’ stress levels, health issues, and productivity.
However, researchers are facing two main challenges. On the one hand, there is a lack of
empirical evidence of the impacts of the interior environment on human experience and a method
to quantify the impacts. On the other hand, there is no applicable laboratory setting since it is
impractical to change design features after construction.
Neuroscience evolved as a new method to differentiate occupant’s experience and
perception in an indoor environment. The study of the Autonomic Nervous System (ANS) and
emotion in the psychophysiological domain built the foundation for quantifying occupants’
experience based on their biometric data. Furthermore, virtual reality has a long history of
application in the architecture and psychology fields. VR can be applied as experimental apparatus
due to advantages such as highly immersive visualization, environmental manipulation, accurate
measurement of performance. After the experiment, machine learning algorithms were used to
predict user preferences based on design features and biometric data inputs.
2

This chapter introduces indoor environment quality (IEQ), neuroscience, virtual reality (VR),
and machine learning. Specifically, it includes relationships between IEQ and building energy
consumption, financial profits, occupants’ well-being, health, and productivity. It demonstrates the
role of neuroscience in architecture and how biometric data can be used to quantify the human
experience. In addition, the applications of virtual reality (VR) in architecture and psychology are
discussed. Last but not least, this chapter explains machine learning and its applications in
architecture.
1.1 Office indoor environment and occupant productivity
According to Centers for Disease Control and Prevention (CDC), indoor environment
quality (IEQ) refers to the quality of a building’s environment that is associated with the health and
well-being of occupants living in it. People spend more than 90% of their time indoor (U.S. EPA
n.d.), thus it is important to understand how IEQ affects building energy, occupants’ health, well-
being, and productivity.
1.1.1 IEQ and Building Energy Cost
Indoor environment quality (IEQ) and occupants’ comfort are essential to building
performance.
IEQ is one of the categories in the green building rating system: the Leadership in Energy
and Environmental Design (LEED). Although the IEQ category reflects awareness of the
importance of occupants’ comfort and productivity, the LEED mainly focuses on site, building
energy use, water, materials, and so forth. Several studies have provided evidence that occupants did
not report higher satisfaction levels in green buildings compared to the experience in conventional
buildings (Paul and Taylor 2008; Altomonte and Schiavon 2013). Many green building rating systems
such as LEED are devoid of direct attention to occupants’ comfort and productivity (Al Horr et al.
3

2016). Consequently, energy use will be increased since the energy consumption in the operation
phase mainly depends on criteria for the indoor environment (Sarbu and Sebarchievici 2013).
Occupants take actions to adapt the indoor environment to their desired comfort levels in terms of
interaction with the control system of the building, such as opening or closing windows, switching
or dimming lighting, increasing or decreasing thermostats (D’Oca, Hong, and Langevin 2018). Thus,
a poor indoor environment results in increased energy expenditure and costs in the operation phase.
1.1.2 IEQ and Financial Benefits
Many studies have reported that increasing investment in a better indoor environment will
lead to substantial financial benefits for organizations and companies, and thus boosting economy.
In developed countries, the salaries of office workers exceed many times the cost of
operational cost of a building (Kosonen and Tan 2004). Therefore, the profits gained by a marginal
improvement in performance will transcend the initial investment in the building environment. The
estimated financial benefits of improving indoor environments exceed costs by a significant factor
(18 – 47)(W. J. Fisk 1997). An annual economic benefit of $10 billion to $30 billion was predicted
when a 20-50% reduction in SBS can be achieved (W. Fisk 2002). The improved overall indoor
working environment can result in a 0.5% to 5% productivity gain, and the corresponding annual
productivity gain is $20 billion to $200 billion for U.S. office workers (W. Fisk 2002). A nationwide
survey-based study implied that a 20% improvement in performance could be reached by a better
indoor environment, equivalent to £135 bn per year (D. Clements-Croome 2015). These estimated
benefits can be a strong incentive for stakeholders to provide a better indoor environment for
occupants.

4

1.1.3 IEQ and Human Well-being
Four main factors that affect IEQ are defined: thermal comfort, indoor air quality, acoustics,
and visual comfort (Sarbu and Sebarchievici 2013). Thermal comfort is defined as occupants’
satisfaction level with thermal conditions (Al Horr et al. 2016). The perception of thermal comfort
depends on various factors such as clothing, activity, age, sex, metabolism rate, and so on (Sarbu and
Sebarchievici 2013). Thermal comfort has considerable impact on occupants’ productivity because
dissatisfaction with thermal environment induce absenteeism and low productivity (W. J. Fisk
1997)(Lan, Wargocki, and Lian 2011). Indoor air quality simply refers to the quality of the indoor.
The judgement of air quality can be very complex, since it is dependent on a broad range of factors
such as temperature, ventilation rate, smell, air contaminants (Al Horr et al. 2016). Acoustics, in this
context, refers to noise in indoor environment. According to Al Horr et al. (2016), noises can be
attributed to two sources: external environment and internal environment. Noises from external
environment are caused by traffics, the public, and machinery, whereas internal noises come from
conversations, electric equipment, etc. Similarly, too much undesirable noises will compromise
occupants’ working performance (Balazova et al. 2008). Moreover, visual comfort links to a number
of factors such lighting, aesthetics, views, and spatial perception (Al Horr et al. 2016). Visual
comfort is also important because most of office tasks rely on vision.
A great number of studies have validated the significance of IEQ factors and their impact on
occupants’ health and well-being. Fisk (1997) concluded four links between health and productivity
and the IEQ: acute respiratory diseases (ARIs), allergies and asthma symptoms, Sick Building
Syndrome (SBS), and direct impacts of indoor environments on occupants’ performance. According
to Fisk (2002), ARIs such as influenza and common colds are related to ventilation rates, occupancy
density, space sharing, and so forth. Allergies and asthma can be linked to a variety of indoor
characteristics, including indoor allergen concentrations, moisture, mold problems, pets, and
5

tobacco smoking. SBS refers to occupants’ discomfort or even acute disease attributed to an
unhealthy indoor environment (Boubekri 2008). Redlich, Sparer, and Cullen (1997) ascribed the
source of SBS to four factors: Air contaminants such as volatile organic compounds (VOCs),
mechanical ventilation, psychological factors such as job satisfaction and stress, and host factors
including sex, atopy, pre-existing disease, and so on. They added that although “Objective
physiological abnormalities and permanent sequelae” are hardly caused by SBS, SBS provokes
uncomfortable and disruptive symptoms, compromising occupants’ productivity (Redlich, Sparer,
and Cullen 1997). Therefore, IEQ factors that fail to fulfill occupants’ needs will provoke healthy
issues.
1.1.4 IEQ and occupants’ productivity
Productivity is defined as a ratio of input to output (Sink 1985). In an office context,
productivity refer to ratio of company turnover to employee cost (Al Horr et al. 2016). Individual
productivity is essential for an organization to meet its goals. Productivity takes into account more
than just how much money an employee is making for the company.
In an organizational working environment, personal, social, organizational, and
environmental factors are identified as four main aspects that affect individual productivity (Al Horr
et al. 2016). Occupants’ productivity will be compromised if they are suffering from stress due to
dissatisfaction with any aspects. The stress level is determined by additional factors: individual
coping skills that result in variability between individuals and time workers spent in this environment
(Rashid and Zimring 2008). Rashid and Zimring (2008) provided evidence that environmental
design plays a vital role in stress. Two categories determine the relationship between the physical
environment of a building and individual stress: Indoor environmental variables and interior design
variables (Rashid and Zimring 2008). Plenty of other studies have echoed that a well-designed
6

indoor environment boosts occupants’ performance, whereas a poor indoor environment hinders
performance. For instance, a survey study shows that approximately two-thirds of the occupants
claimed it could improve 10% or more productivity by providing a better indoor environment
quality (D. J. Clements-Croome et al. 2000).
1.1.5 Factors that affect comfort and productivity
Al Horr et al. (2016) summarized eight environmental factors in the office context that
significantly impact employees’ productivity. These factors are “indoor air quality and ventilation,
thermal comfort, lighting and daylighting, noise and acoustics, office layout, biophilia, and views,
look and feel, location and amenities“(Al Horr et al. 2016).
Indoor air quality and ventilation relate to the quality air indoor and exchange of air between
indoor space and outdoor space. Thermal comfort describes occupants’ satisfaction level of thermal
conditions. The judgement of thermal comfort is not limited to air temperature, it also can be
influenced by air humidity, air speed, and so on. Lighting and daylighting, in this context, refers to
artificial lighting in office environment and sunlight received indoor. Noise and acoustics indicate
sound levels in an office environment. It is a combination of sounds from external sources (e.g.,
public and traffics) and sounds from internal sources (e.g., conversations and office equipment).
Office layout denotes the arrangement of office plan. It is often associated with density of space,
privacy of individual working environment, access to programs and core area. Biophilia means that
plants and living creatures appeal to human naturally. In this context, it means occupants’ preference
of having plants in their working environments. Views refers to scene of nature or city from
windows. Look and feel contains texture and color of surfaces of indoor environment, design
aesthetics, spatial shapes such as ceiling height and openness. Location and amenities refer to
7

location of office and services that office can access. They are factors not physically included in an
office indoor space, but they are critical to the quality of working environment.
1.2 Neuroscience and Architecture
This chapter includes definition of neuroscience, and why it is important to architects. In
addition, this chapter illustrates fundamental concepts of autonomic nervous systems. Finally, there
are applications of neuroscience in architecture.
1.2.1 Neuroscience as an emerging tool to quantify human experience
The role of buildings is beyond physical and functional. As Peter Buchanan pointed out,
people often fail to consider psychological and cultural aspects of sustainability. Instead, they focus
more on ecology and technology. (Buchanan 2012). However, there is a lack of measures of
psychology and culture unless post-occupancy evaluations are conducted (Sarah Robinson 2015). In
Digital Era, efforts were taken to quantify building performance, but little value was placed on
quantifying individual experience. Whether consciously or not, people’s mind and behavior are being
changed by architecture design. “architecture design changes our brain and our behavior” (Robinson
and Pallasmaa 2015).
It is the ’architect’s’ responsibility to understand the impacts of the built environment on
human mental function and behavior (Robinson and Pallasmaa 2015). However, one challenge
architects face is the lack of empirical evidence of the impacts of interior built environment on
human experience and the method to quantify how human experience is affected by architectural
design features (Ergan, Shi, and Yu 2018). Traditional methods, asking occupants’ evaluation of the
space configuration in terms of self-reports, have been unpractical and inaccurate. On the one hand,
8

there are limitations to changing space configuration after construction. On the other hand, self-
reports suffer from a range of sources of imprecision (Paulhus and Vazire 2007).
To understand how long-term mental health and performance can be affected by working
environments, architects started to seek solutions in neuroscience domain. Neuroscience has the
potential to fill in the vacancy of architects’ intuitive design with robust evidence.
1.2.2 Neuroscience and Autonomic Nervous System
Neuroscience is a field that understands the mind and behavior by measuring the activities in
the nervous system. The autonomic nervous system is one branch of nervous system, and it is in
charge of bodily activities that cannot be controlled consciously.
Neuroscience has a broad collection of disciplines: biology, experimental psychology,
cognitive science, chemistry, anatomy, physiology, and computer science that study the relationship
between brain and behavior, which forms human sensation, perception, cognition, memory, and
emotion (Kandel, E. R. et al. 2000). Psychology studies the human mind and behavior, while
neuroscience aims at understanding how the mind and behavior are affected by the brain and body.
In order to understand the physiology, psychology, and emotion of humans, it is critical to
review the nervous system, which is the central controlling, regulatory, and communicating system
of the human body. The autonomic nervous system (ANS) is part of the nervous system that
controls and regulates bodily functions that are unconsciously directed, such as muscles of vessels,
the digestive system, cardiac muscle, sweat, lacrimal glands, and so forth (Izadyar 2017). The ANS
has three branches: sympathetic, parasympathetic, and enteric (Izadyar 2017). The sympathetic and
parasympathetic branches work as “opposite” roles, in which one system activates, and the other
inhibits a physiological response (Figure 1-1)(Izadyar 2017).
9

Figure 1-1 Roles of the sympathetic and parasympathetic systems. Redrawn from (Jänig 1989).
1.2.3 Quantify human experience through biometric sensors
Human experience arises from interaction with surrounding environment, causing subjective
sensation and feelings. There is a lack of methods to quantify human experience. In architecture
fields, it is common to use semantic differential scales to quantify, which is not reliable. Biometric
sensors are wearable sensors that can detect human physiological data.
Researchers have built up an empirical understanding of how design features influence
human mental function and behavior (Robinson and Pallasmaa 2015). People appraise their
surrounding environments through a series of stimuli triggered by “materials selected, spatial
relations, formal proportions, scale, patterns, rhythms, tactile values, and creative intentions” (Sarah
Robinson 2015). A building environment can be designed to activate the sympathetic nervous
10

system, causing the metabolic system’s arousal. Alternatively, it can bring relaxation and comfort to
occupants, which are controlled by parasympathetic nervous systems.
To quantify the physiological responses and experiences triggered by the sympathetic and
parasympathetic nervous system, researchers have been integrating a collection of biometric sensors
in their studies. The two essential features of ANS identified by Levenson, coherence and specificity,
built the foundation for identifying emotion through biometric data (Levenson 2014). Several
studies have investigated the relationship between a certain design feature (e.g., daylight, illuminance
level) in the built environment and human experience by biometric sensors (Ergan et al. 2019;
Radwan and Ergan 2017; K Chamilothori et al. 2019). However, the empirical study of the impact of
a collective set of design features on human experience is still insufficient (Ergan et al. 2019).
1.3 Virtual Reality
Due to the flexibility of virtual reality (VR) technologies to build controllable environments
with immersive and realistic experiences, VR started to spread into fields of both psychology and
architecture decades ago. VR as an experimental tool is acknowledged and even recommended by
scholars in psychology and architecture research.
1.3.1 What is Virtual Reality?
Extended reality (XR) consists of augmented reality (AR), virtual reality (VR), and mixed
reality (MR).
Augmented reality is an interactive experience that superimpose digital information onto real
world objects. It can be experience though various devices such as phone, pad, and AR glasses. It is
different from virtual reality. Virtual reality is an immersive virtual world created by computers.
11

Users can enter this virtual world by wearing VR headsets. It is completely digital. Mixed reality is
term that combines virtual reality, augmented reality, and reality.
Milgram and other researchers introduced the concept of the “virtuality continuum“;
accordingly, mixed reality refers to any points on the spectrum of the real environment and virtual
environment (Figure 1-2) (Milgram et al. 1995). Therefore, AR, VR, and MR are not monolithic
terms or simplistic concepts (Greengard 2019), and they can be conveyed through various tools and
devices in terms of any combinations.

Figure 1-2 reality virtuality continuum. Redrawn from (Milgram et al. 1995).
Virtual reality (VR) is commonly referred to as a three-dimensional, interactive, and
computer-generated environment (Greengard 2019). Today, common VR platforms can be
classified into three paradigms: 1) head-based: The head-mounted display (HMD) is the most
recognizable device (Figure 1-3). 2) stationary: the VR device is fixed, and the projectors are used to
create a virtual experience (Figure 1-4). 3) hand-based: it is based on smartphones or tablets (Figure
1-5)(Zhang et al. 2020).
12

Figure 1-3 VR Head-mounted Display, Dec 31, 2019, Insider, (Insider 2019)

Figure 1-4 Semi-immersive VR, n.d., visbox, (Visbox n.d.)
13

Figure 1-5 VR on smartphone, 2017, (Bnext3D 2017)
1.3.2 A brief history of Virtual Reality
The development of XR can be traced back to the Leicester Square panorama in the 1780s.
It was a view of the Cities of London and Westminster painted on circular structure, delivering a
sense of presence in the scene to viewers. The Cave Automatic Virtual Environment (CAVE)
introduced by the Electronic Visualization Laboratory (EVL) in 1992 was considered the first truly
immersive virtual experience (Greengard 2019). The significance of CAVE lies in the fact that it
conquered the challenge that early HMDs are chunky, and the applications were restricted
(Greengard 2019). In 1994, the second generation of the CAVE was created, and many researchers
started to apply the use of VR in various fields, such as architecture, education, gaming, engineering,
information visualization. (Greengard 2019). The evolution of VR technology never stopped
marching. From the prototype of the Oculus Rift in 2011 to the Vive headset from HTC, the
PlayStation VR, the Cardboard released by Google, VR technology became more prevalent due to
its growing availability, portability, and affordability (Zhang et al. 2020).

14

1.3.3 The use of Virtual Reality in psychology study
VR has been widely accepted as a new research tool in psychology and clinical (Foreman N
2006; Gregg and Tarrier 2007; C. J. Wilson and Soranzo 2015). Increasing use of VR technologies
can be found in prominent areas of psychology such as perception, memory, problem-solving,
mental imagery, and attention (C. J. Wilson and Soranzo 2015; Gaggioli 2001). The VR application is
acknowledged and even encouraged by researchers as an experimental apparatus (Gaggioli 2001).
VR is used to generate complex scenarios in the clinical and mental health fields. Many studies have
employed VR in the treatment of specific phobias such as fear of flying, fear of height, and fear of
spiders (Krijn et al. 2004; Carlin, Hoffman, and Weghorst 1997; Gregg and Tarrier 2007). Other VR
applications have been found in cognitive assessment and rehabilitation of patients with traumatic
brain injury, stroke, and dementia (Lee et al. 2003; Gregg and Tarrier 2007). Moreover, VR has been
developed to treat PTSD in war veterans (Gerardi et al. 2008).
The advantages of using VR in psychology can be concluded as 1) stimuli are presented in an
immersive and three-dimensional space; 2) researchers have greater control over the stimuli; 3) it
allows combinations of multimodel sensory input simultaneously such as audio, haptic, olfactory and
motion; 4) it is possible to evaluate complex behaviors; 5) it permits to create an experimental scene
that might be impractical, risky, or ethically questionable in real life (C. J. Wilson and Soranzo 2015).
Despite the potential of VR technologies being applied in psychology studies, researchers
must take into account the fundamental limitations. First, considerable efforts have to be devoted to
making participants feel “real.’ Next, VR-induced side effects make VR inaccessible to certain
studies and subjects. Most importantly, whether experiment results of psychology study obtained in
VR can be extended to real-life (Foreman N 2006; Gaggioli 2001).

15

1.3.4 Bring Virtual Reality into the Architecture
VR has been increasingly adopted in the AEC industry. VR has been employed to improve
efficiency and productivity in design, decision-making, education and training, and construction
safety management (Alizadehsalehi, Hadavi, and Huang 2020). Although there were still problems of
interoperability of VR data and BIM, VR was identified as a key solution to leverage the full
potential of Building Information Modeling (BIM) (Alizadehsalehi, Hadavi, and Huang 2020).
Architects have applied VR in the creation and perception of design. It is challenging to
accommodate stakeholders’ knowledge and needs in the design through a traditional approach
because they lack professional engineering knowledge to read 2D drawings and 3D modeling (Lin et
al. 2018). VR can incorporate the stakeholders into the early design phase, improving the
communication between the stakeholders and the design team. In addition, one study provided
evidence that VR can improve the engaging experience in collaborative design and facilitate
conceptual design (Figure 1-6)(Koutsabasis et al. 2012).
Furthermore, VR technology has been adopted in human-building interaction (HBI) (Zhang
et al. 2020). HBI studies can be classified as 1) wayfinding behavior and spatial perception; 2)
emergency spatial perception; 3) energy-consumption-related behavior (Zhang et al. 2020). VR was
adopted to investigate the impact of interior design (Radwan and Ergan 2017; Ergan et al. 2019) and
façade pattern (K Chamilothori et al. 2019) on human experience and psychological responses. The
VR technology was recognized as superior to traditional methods because of the controllable
environments, immersive and realistic experience, higher ecological validity, and fewer ethical
problems involved in human study experiments (Zhang et al. 2020).
16

Figure 1-6 VR in architecture, June 20, 2020, ARK Architects, (ARK architecture 2020)
1.4 Machine Learning
This chapter includes machine learning and supervised learning methods. Machine learning
started to bloom over the past two decades.
Machine learning is computer algorithm that learn through data and improve itself through
experience.
The advancements of three key technologies accelerate the application of machine learning
in various fields: First, the rapid development of sensing and Internet of Things (IoT) technologies
give rise to an exponential growth of data. IoT refers toa system that gathers information from
objects via a wireless network. For example, IoT devices can be a smartphone, smartwatch, smart
refrigerators, sensors, and so on; second, GPU (Graphics Processing Units) and TPUs (Tensor
Processing Units)) provide more affordable and efficient computational resources. A computer uses
GPU to render images, videos, and games. TPU was developed by Google to accelerate machine
17

learning computation, and more solid machine learning algorithms have been developed and
improved (Hong et al. 2020).
1.4.1 What is Machine Learning?
Machine learning is a sub-area of artificial intelligence (AI). The central goal of AI is to
mimic human intelligence to perform tasks. Machine learning integrates computer science, statistics,
neuroscience, and other disciplines involving automation and decision-making under vagueness
(Figure 1-7)(Jordan and Mitchell 2015). More specifically, the machine-learning is defined as
exploring progress to optimize the performance metrics through a series of candidate programs
(Jordan and Mitchell 2015).
A well-known example of machine learning is image recognition. For instance, a machine
learning algorithm can train a computer to identify images of dogs among other pictures. Some
people may take this task for granted. A substantial amount of information is processed and
organized in the human brain to recognize a dog’s picture. With the help of machine learning, a
computer can mimic the way a human does. The algorithm dissects a dog image into pixels and
corresponding color values, along with the " dog " label. After watching tens of thousands of
pictures of dogs and other things, the computer will be able to find dog pictures among new pictures
with a certain precision. The strategies to further improve prediction precision are increasing the
number of data sets, tuning hyperparameters (parameters that control the algorithm), finding more
suitable algorithms, and better feature extractions (e.g., removing background noise before training).
In an era of “big data.”, the rapid development of computers, mobile, and networks caused
exponential data growth. Data scientists take advantage of the machine learning to gain insights,
predictions, and solutions from a large amount of data (Géron 2017). What makes a machine
learning solution more compelling is that machine learning can leverage the granular and
18

personalized nature of the data (Jordan and Mitchell 2015). Personal mobile and computer allow
data to be collected individually, and machine learning then can take advantage of the individual data
and provide customized solutions. Today, machine learning applications can be commonly found
across the fields of commerce, biology, astronomy, neuroscience, and other data-intensive empirical
sciences (Jordan and Mitchell 2015).

Figure 1-7 Machine learning technologies, Jan 11, 2022, EURIX, (EURIX 2022)
In recent years, machine learning has been widely applied in architectural research
throughout the whole building life cycle (Hong et al. 2020). Across the building life cycle, building
performance simulation and building control are the most trending research area (Hong et al. 2020).
Machine learning enables analysts to acquire performance simulation within a shorter time and with
higher accuracy (Singaravel, Suykens, and Geyer 2018). Machine learning in building control
provides minimized energy costs without compromising user comforts (Figure 1-8) (Hong et al.
2020).
19

Figure 1-8 machine learning for smart building applications (Djenouri et al. 2019)
In general, algorithms convert machine learning problems into a function composed of
inputs and outputs, and the learning process improves the accuracy of pairing inputs and outputs
(Jordan and Mitchell 2015).
1.4.2 Supervised Learning Method
Machine learning contains four types of algorithms: supervised learning, unsupervised
learning, reinforcement learning, and semi-supervised learning (Géron 2017). Supervised learning
makes machine learn under supervision of human beings. Human tells machine which is the correct
answer. On the contrary, unsupervised learning lets machine learn on its own.
Supervised learning is the most extensively used machine learning method (Jordan and
Mitchell 2015). Supervised learning takes in a group of pairs (x, y), representing predictors and
labels. The goal is to predict y^ based on input x. Both input and output can be taken in various
forms, such as vectors, images, audio. (Jordan and Mitchell 2015). A supervised learning task can be
divided into regression and classification (Géron 2017). A typical regression task is to predict
numerical value, whereas a classification task predicts the labels (e.g., “spam,” “not spam”). The
classification focus on various problems such as simple binary classification, multiclass classification
20

(exclusive classes), multilabel classification, ranking problems, and so forth (Géron 2017). The
mapping function f(x) that takes in x and y has numerous forms such as logistic regression, decision
trees, support vector machines, neural networks. Different architectures and algorithms with
different mathematical structures meet the diverse needs of data scientists, offering trade-offs
between computational complexity, the size of data, and performance (Jordan and Mitchell 2015).
On the contrary, unsupervised learning trains unlabeled data (Géron 2017). The principal
unsupervised learning algorithms can be used to solve dimensionality reduction problems, clustering
problems, and anomaly detection. (Géron 2017).
1.5 Summary
This chapter discussed indoor office environment, neuroscience, virtual reality, and machine
learning. In particular, it stressed that indoor office environment have impacts on building energy
use, financial benefits, occupants’ health, well-being, and productivity. Next, the definitions of
neuroscience, autonomic nervous system, and the potential method to quantify human experience
were explained. Furthermore, it introduced the definition and history of virtual reality, the
applications of virtual reality in architecture and psychology. Finally, the chapter introduced machine
learning and two types of commonly used algorithms: supervised learning and unsupervised
learning.

21

Chapter 2 Background and literature review
This chapter includes three design features that impact occupants’ well-being and
productivity: daylight and lighting, biophilia and view, look and feel. Next, it illustrates that
measuring the activities of the autonomic nervous system is the key to the differentiation of
emotions and the connections between heart rate, heart rate variability, skin temperature,
electrodermal activities, and the autonomic nervous system. It then introduces the ecological validity
of virtual reality, presence, immersion, and research in architecture concerning presence.
2.1 Design Features that Affect Occupant Well-being and Productivity
Three categories, daylight and lighting, views and biophilia, look and feel, are concluded.
They are subdivided into daylight, lighting, biophilia and views, color, ceiling height, and texture
(Table 2-1). The principal conclusions regarding the design features in research are listed as well as
the preferred variables of the feature.
Table 2-1 Literature review of design features and their impact on occupants
Design Features Variables Reference Conclusion
daylight Presence of daylight (Boubekri 2008;
Ander 2003;
Galasiu and Veitch
2006)
daylight is essential in the
working environment;
occupants prefer daylight
over artificial lighting;
lighting 50% higher than 500lux in
standards; 800 lux; preferred
illuminance increases when
maximum available
illuminance range increased
(337, 523, 645 lux)
(Boubekri 2008;
Logadóttir,
Christoffersen, and
Fotios 2011)
Workers prefer higher
levels than those
recommended by the
professional organization.
Biophilia & views presence of plants in office;
nature views.
(Gray and Birrell
2014; I. Elzeyadi
2011; Africa and
Sachs 2016)
Incorporating biophilic
design into the office boost
productivity and promote
workplace satisfaction.
22

Color Users preferred white room
over red and green room; in
one experiment, users most
like beige and white office,
least like orange and purple
office; red office cause lower
satisfaction level; red office
induces anxiety and stress
(Nancy Kwallek,
Lewis, and
Robbins 1988;
Nancy Kwallek
1996; Mahnke
1996)
Color affects occupants’
moods and performance.
Ceiling Height Preference for the ceiling
height higher than 6ft (1.83m)
to a peak at 10 ft (3.04m) and
decrease after.
(Baird, Cassidy,
and Kurr 1978;
Vartanian et al.
2015)
Rooms with higher ceilings
were more likely to be
judged as beautiful.
Texture (walls and
floor)
No study shows preference in
an office context
(Sadalla and Sheets
1993)
People have different
psychological responses to
different materials.

2.1.1 Daylight and Lighting
Daylight is one of the indispensable elements of human life. Historically, the awareness of
the importance of sunlight was reflected in great architects’ works: Le Corbusier, Frank Lloyd
Wright, Mies van der Rohe, etc. (Boubekri 2008). Maximizing daylight indoors according to
geological location is widely accepted as one of the major principles for architects. In addition, the
need for artificial lights is decreased because of sufficient daylight, resulting in less building energy
consumption and carbon emission. Daylight also induces a conducive impact on occupants’
physiological and psychological health, forming a working environment that increases workers’ well-
being and productivity (Boubekri 2008).
Electrical lighting accounts for 25%-40% of total electrical energy use for a typical US
commercial building (Krarti 2020). The three main categories that are responsible for electricity
generation are fossil fuels (coal, natural gas, and petroleum), nuclear energy, and renewable energy
(Energy Information Administration 2021). The burning of coal, natural gas, and oil for electricity is
the biggest contributor to global greenhouse gas (GHG) emissions (Edenhofer et al. 2014).
Increasing the use of daylight in a building is a great strategy to reduce energy use and mitigate GHG
23

emissions. A number of studies have reported that the building can reach up to 60% of energy-
saving by incorporating daylight controls (Ihm, Nemri, and Krarti 2009)(Doulos, Tsangrassoulis, and
Topalis 2008).
Daylight affects human bodies in two ways: tuning our metabolism, endocrine system, and
hormone systems and producing vitamin D through photosynthesis on our skin (Boubekri 2008).
An area in the brain controls most metabolic processes, endocrine, and hormone systems called the
hypothalamus, which is correlated with the circadian cycle, breathing, emotional balance,
reproduction, heat regulation, etc. (Boubekri 2008). The so-called body block exists in this area, and
it contributes to human’s daily activity and sleep rhythms (Boubekri 2008). Daylight acts as a
principal external cue for the body clock (Boubekri 2008). The scarcity of daylight or insufficient
exposure time of daylight will provoke an internal disturbance, resulting in drowsiness, social
anxiety, insomnia, bulimia, and even internal cancers (Boubekri 2008). Furthermore, daylight is vital
for the formation of vitamin D. Vitamin D plays an important role in bone development and
growth. 80%-100 % of vitamin D is generated in terms of photosynthesis through human skin
(Glerup et al. 2000). Inadequate absorption of daylight will cause a number of health problems such
as bone thinning, cancers, high blood pressure, depression, and immune-system disorders (Boubekri
2008).
Many studies have shown that lighting conditions have an effect on occupants’ mood,
therefore boosting or suppressing productivity. Artificial lighting can never achieve the lighting
quality and dynamic dimension of daylight indoor (Boubekri 2008). Therefore, the majority of
workers prefer daylight over artificial lighting and think daylight is a superior lighting source (Galasiu
and Veitch 2006). One study attributed more sick leave hours to the poor rating of daylight and
poor view in the office (Ihab Elzeyadi 2011). Good lighting induces excitement, alertness, and
24

dominance, whereas poor lighting provokes boredom, dullness, and submissiveness (Boubekri
2008).
2.1.2 Biophilia and views
Wilson defined biophilia as “the innate tendency to focus on life and lifelike processes” (E.
O. Wilson 2018). Many studies have verified the positive relationships between human well-being
and natural environments. One study provided evidence that participants reported significantly
higher satisfaction levels in all green or natural habitats than those in urban environments
(MacKerron and Mourato 2013). Unfortunately, rapid urbanization brings about vast transformation
to our built environments, resulting in fewer opportunities for people to access natural
environments in an urban region.
Bringing natural features into the building or introducing natural views through windows will
increase occupants’ connections to natural environments. The insertion of plants into a building will
improve indoor air quality and raise comfort, satisfaction, well-being, and performance (Gray and
Birrell 2014; Africa and Sachs 2016). Occupants preserve strong physical and psychological needs
for exterior views, especially when the views contain natural features and vegetation, which boosts
occupants’ comfort and working performance (Africa and Sachs 2016). Elzeyadi conducted
questionnaires among 175 employees and found out that poor rating of daylight quality and poor
views increased the more uses of sick leave hours (I. Elzeyadi 2011).
Biometric signals and psychology have been incorporated into studies to provide more
robust evidence of the benefits of plants and nature views. According to Chang and Chen, who
employed electromyography (EMG), electroencephalography (EEG), and blood volume pulse
(BVP) in their study, occupants suffered from a higher degree of stress and anxiety in the room
without the presence of plants and exterior views (Chang and Chen 2005). In another study that
25

monitors blood pressure and emotion, participants reported higher productivity and perceived lower
stress in the room with plants. (Lohr, Pearson-Mims, and Goodwin 1996)
In summary, biophilia refers to people’s inherent needs to affiliate with natural habitats. The
insertion of plants into the office and views through windows help build a high-quality indoor
environment, resulting in improved work performance, high satisfaction levels, lower stress, and
comfort.
2.1.3 Look and feel
Color is among the most common and most prominent stimuli people receive from the
environment. People tend to connect colors to objects or physical spaces; for instance, they may
associate green and blue with nature and the sky (Naz and Epps 2004). Studies show that the color
of the space influences human perception of the space, further affects human performance, mood,
and well-being (Nancy Kwallek, Lewis, and Robbins 1988; Mahnke 1996; Naz and Epps 2004).
Despite the discrepancy among male and female groups, one study demonstrated that the saturation
of color might influence people’s moods, such as depression, confusion, and anger (N Kwallek et al.
1996). In addition, subjects reported they prefer to work in the beige and white offices, while they
least like to work in the orange and purple color offices (N Kwallek et al. 1996). Color influences
occupants’ emotions and tweaks occupants’ perception of the volume, weight, temperature, time,
and noise (Mahnke 1996).
Texture, in this case, denotes the material of the interior surface. The textures and color
jointly trigger people’s memory and therefore affect occupants’ moods and feelings (Al Horr et al.
2016). Brick, for example, elicits feelings of durable, heavy, and hard, whereas people sense wood as
a softer, lighter, warm, and organic material (Sadalla and Sheets 1993).
26

Spatial shapes (such as openness, ceiling height, symmetry) form people’s perception of the
space. Ceiling height and perceived closure affect aesthetic judgments of the space (Vartanian et al.
2015). People tend to prefer ceilings that are higher than the height in the building code, and they
are willing to pay for the high extra cost of adding ceiling height (Vartanian et al. 2015). The
openness of the space not only affects beauty judgments but also has an impact on performance and
job satisfaction (de Croon et al. 2005; Vartanian et al. 2015).
2.2 Physiological background
The Autonomic Nervous System (ANS) plays an essential role in emotion, whether it is
generation, expression, experience, or recognition of emotion (Levenson 2014). In neuroscience,
researchers have been using statistical analysis of physiological signals from the ANS to differentiate
emotional states.
2.2.1 How do humans sense emotions?
Levenson provided evolutionary/functionalist models from a “bottom-up” view concerning
the correlations between ANS and emotion (Levenson 2014). In this view, subjective experience of
emotion and cognitive functioning are shaped by ANS and other peripheral responses (facial
expression, vocalization, and motor programs) (Levenson 1999; 2011; 2014). The emotion elicitation
model proposed by Levenson illustrated the role of ANS in emotion (Figure 2-1). When the brain
centers detect sensory inputs, the emotions will be activated and further create activities of ANS and
between other response systems. The pattern of activity within the ANS (e.g., cardiac, vascular,
electrodermal) and those between other response systems triggered by emotion help the
organization prepare for the responses (Levenson et al. 2016).
27

Figure 2-1 Emotion elicitation model, (Levenson 2014)
Two critical features, coherence and specificity, are identified in the activity of ANS.
Coherence refers to the activities between ANS and the other response systems are coordinated and
regulated by emotions (Levenson 2014). Specificity, on the other hand, hypothesizes that particular
emotion is associated with patterned activities of ANS (Levenson 2014). Specificity paved the way to
identify a certain emotion through measuring biometric data since different patterns of ANS activity
are related to different emotions.
2.2.2 Measure ANS
Most research adopted measures of heart rate and skin conductance to quantify ANS
activities, thus pinning down the emotion (Kreibig 2010). Heart rate and skin conductance are also
correlated with the 2-dimensional model that includes valence (negative-positive) and arousal (low-
high). The two-dimension model of valence and arousal form a circumplex with four quadrants
28

(Figure 2-2) (James A. Russel 1980). This model defines an emotional state by the coordinates rather
than a discrete label.

Figure 2-2 Two dimensions of valence and arousal. (Wang, Nie, and Lu 2014)
Valence refers to the differentiation of emotion (e.g., stressed-relaxed, happiness-sadness) in
the positive to negative scale, whereas arousal demonstrates the intensity of the emotion (James A.
Russel 1980; Cacioppo, Tassinary, and Berntson 2016; Kreibig 2010). Studies provided evidence that
arousal is connected to an increase in electrodermal responses (Bradley and Lang 2000; Gomez and
Danuser 2004; Levenson et al. 2016; Vrana and Rollock 2002). Other studies indicated that valence
is more associated with heart rate, and heart rate acceleration and deceleration correspond with
positive stimuli and negative stimuli, respectively (LANG et al. 1993; Palomba, Angrilli, and Mini
1997; Levenson et al. 2016).
2.2.3 Heart rate and heart rate variability
Heart rate has been the most common measurement in psychological studies due to its
simplicity, inexpensiveness, and reliability (Kreibig 2010; Levenson et al. 2016). In addition, heart
rate variability has been adopted to reveal cardiovascular dynamics and their central and peripheral
29

autonomic control (Levenson et al. 2016). Heart period (in msec), the time between heartbeats, are
usually converted to heart rate (in beats/min or bpm) (Levenson et al. 2016). Measures of HRV can
be categorized into time domain and frequency domain metrics (Levenson et al. 2016). Time-
domain metrics measure the variance among heart periods, whereas the frequency domain metrics
break down variance into frequency bands (Camm et al. 1996; Levenson et al. 2016). The
cardiovascular system is regulated by both the sympathetic and parasympathetic branches of the
autonomic nervous system, which corresponds to the valence of the emotion (Levenson et al. 2016).
2.2.4 Electrodermal activity and skin temperature
Electrodermal activity (EDA) or galvanic skin response (GSR) refers to the variation of the
variation of the electrical properties of the skin caused by sweat secretion (Benedek and Kaernbach
2010). Through the relationships among voltage, current, and resistance in Ohm’s law, EDA can be
determined by measuring skin conductance (Boucsein 2012). It reflects the electrical changes of the
skin due to sweat glands (Kim, Bang, and Kim 2004). As mentioned, EDA only receives signals
from the sympathetic nervous system, and thus it is a good indicator of arousal level in a 2-
dimensional emotion model. Accordingly, measuring EDA cannot discern the type of emotion, but
the intensity of it. Examining electrodermal activity (EDA) in areas of the body where eccrine glands
are located, such as hands’ palmar surface, is a typical way to measure sweating (Levenson et al.
2016). Compared to other sweat glands mainly affected by temperature, eccrine glands are
susceptible to psychological stimulations (Levenson et al. 2016).
The time series of skin conductance can be decomposed into two components: a slowly
changing tonic activity (i.e., skin conductance level, SCL) and a rapid varying phasic activity (i.e., skin
conductance response, SCR) (Benedek and Kaernbach 2010) (Figure 2-3). SCR is often associated
with event-related responses or non-event-related responses. A common procedure to gauge event-
30

related responses to a certain stimulus is measuring the amplitude of the elicited SCRs that is above a
minimum amplitude (e.g., 0.05 μS) within a predefined time window (e.g. 1 – 3s or 1 – 5s) after the
event onset (Benedek and Kaernbach 2010; Boucsein 2012). Comparing to standard peak detection
method (through-to-peak, TTP), different efficient approaches to decompose skin conductance data
were widely employed in research (e.g., Continuous Decomposition Analysis, Discrete
Deconvolution Analysis) (Benedek and Kaernbach 2010)) (Figure 2-4).
Local changes in blood flow rate caused by vascular resistance or arterial blood pressure
result in variations in skin temperature variation (Shusterman and Barnea 1995). Vascular resistance
is mediated by smooth muscle tone, which is controlled by the sympathetic nervous system
(Shusterman and Barnea 1995). Skin temperature variation exhibits ANS activities; therefore, it is
another measure of emotional states (Kim, Bang, and Kim 2004).

Figure 2-3 EDA is composed of Tonic EDA and Phasic SCRs (redrawn from (Benedek and Kaernbach 2010))

Figure 2-4 Skin Conductance data consists of phasic SC (steep change) and Tonic SC (slow change) (analysis of subject 07 in Ledalab)
31

2.3 Virtual Reality
Virtual reality (VR) demonstrates advantages such as highly immersive visualization,
environmental manipulation, allowing researchers measure subjects’ performance accurately. The
advancement and affordability of computer graphics and VR HMD equipment catalyze expanding
use of VR in pilot studies. A number of studies have demonstrated that VR technology has the
potential to be applied as experimental apparatus in psychological study and clinical (Gaggioli 2001;
Krijn et al. 2004; Carlin, Hoffman, and Weghorst 1997; Lee et al. 2003; Gregg and Tarrier 2007; C. J.
Wilson and Soranzo 2015). VR has also shown potential in architecture design and construction,
especially in pre-occupancy and post-occupancy evaluations (Tseng, Giau, and Huang 2017;
Koutsabasis et al. 2012; Alizadehsalehi, Hadavi, and Huang 2020; Zhang et al. 2020). In addition, a
limited number of studies devoted to the intersection of architecture and psychology, or human-
building-interaction (HBI), such as wayfinding behavior and spatial perception, relationships
between interior design and occupants (Radwan and Ergan 2017; Yeom, Choi, and Kang 2019;
Zhang et al. 2020).
Despite the considerable potential that VR has shown in architecture and psychology
studies, applications in real life are not common yet. The issues of interoperability of data, VR-
induced side effects, the reluctance to adopt new technology are widely-agreed factors that inhibit
VR technology from applying in real-life (Foreman N 2006; Gaggioli 2001). However, the most
criticized limitation is the uncertainty of the ecological validity of VR studies.

32

2.3.1 Ecological Validity of VR
Ecological validity generally refers to whether or not an observed result in a laboratory
setting can be generalized into a naturalistic setting (Schmuckler 2001). It is critical since it
determines whether the human experience and responses triggered in the virtual environment are
similar to those triggered in the real world. Therefore, the achievement of ecological validity in VR
study is crucial to the reliability and applicability of the research.
It is widely accepted that VR as a new assessment instrument in psychological study proffers
enhanced ecological validity than traditional methods (Gaggioli 2001; Foreman N 2006; Parsons
2011; 2015). However, scholars argue that it is not fully substantiated that to which extent, study
outcomes in an immersive virtual environment (IVE) setting can be expanded entirely to the real
environment (RE) (Gaggioli 2001; Zou, Li, and Cao 2017; Deniaud et al. 2015; Paljic 2017).
The best way to examine ecological validity is to compare subjects’ behavioral and
psychological differences between scenarios in IVE and RE with identical settings (Deniaud et al.
2015; Zou, Li, and Cao 2017).
In some cases, however, it is inviable and costly to reproduce a corresponding real-life
setting (e.g., evacuation study, studies of the perception of space configuration). Therefore,
researchers proposed that the “presence” can be utilized as an effective measure of ecological
validity (Deniaud et al. 2015; Zou, Li, and Cao 2017). The underlying assumption is that users’
responses in an IVE will be equivalent to responses in RE if the setting of IVE is indistinguishable
from the setting in RE (Loomis, Blascovich, and Beall 1999; Kuliga et al. 2015).
33

2.3.2 Presence, Immersion, and Measure of Presence
“Presence” and “immersion” are terms that describe the depth of users’ “immersive”
experience in immersive virtual environment (IVE). The presence was defined by Slater et al. as “the
extent and capability of participants in a virtual environment to respond to virtual situations and
events as if these were real” (Slater et al. 2009). Presence is a state of consciousness and a
psychological sense (Slater and Wilbur 1997). The immersion, on the other hand, was distinguished
from the presence as a description of a technology (Slater and Wilbur 1997; Sanchez-Vives and
Slater 2005). It delineates the capability of the system to deliver an “inclusive, extensive,
surrounding, and vivid illusion of reality” to users (Slater and Wilbur 1997). Important factors of
immersion include the field of view, the number of sensory systems, rendering quality, the realism of
the scene, matching (or body tracking), the framerate, and the latency (Slater and Wilbur 1997; Slater
et al. 2009; Sanchez-Vives and Slater 2005).
Many scholars attempt to discover the necessary conditions that contribute to a high-level
presence. According to Diemer et al., users’ judgment of the presence of the virtual environment
(VE) was based on immersion and the degree of arousal they feel (Diemer et al. 2015). This was
echoed with hypotheses derived from Slater and Wilbur that presence is determined by immersion
and users’ subjective evaluation of their degree of “being there” (Slater and Wilbur 1997). In
addition, they proposed that presence is an increasing function of all factors of immersion (Slater
and Wilbur 1997). The positive effects of framerate, body tracking, resolution of displays, haptic
feedback, etc., on presence have been examined (Barfield and Hendrix 1995; Paljic 2017; Sanchez-
Vives and Slater 2005).
There are three approaches to measure the presence. First, the most common approach is
using a questionnaire with ordinal scales (Slater et al. 2009). However, questionnaire-based
34

evaluation was inaccurate and can be affected by prior information (Freeman et al. 1999). The
second approach of measuring is behavioral. The idea is that if users behave the same in immersive
virtual environment (IVE) as in real environment, the level of presence is high (Slater et al. 2009). A
specialization of the behavioral approach is measuring the presence in terms of physiological
measures, such as electrodermal activity, heart rate (Slater et al. 2009). Notably, the difference of
physiological measures between RE and IVE is less evident if the IVE is in mundane situations such
as virtual rooms compared to noticeable difference in building evacuation studies (Figure 2-5) (Slater
et al. 2009). The last approach is the measure of breaks in the presence (BIPs), which refers to
experience bringing users’ awareness to the real-world (e.g., bumping into a wall, getting trapped in
cables) (Slater et al. 2009).

Figure 2-5 Virtual building evacuation study (Zou, Li, and Cao 2017)
2.3.3 Studies of presence in the architectural domain
VR provides human-environment interaction, controllable immersive virtual environment,
and behavior measurement, making VR a potential tool to apply in many domains. So far, VR is
widely implemented as a research tool in experimental settings rather than in real life because the
ecological validity of VR remains unexamined. Ecological validity determines whether or not the
study can be generalized into real life. Several studies have investigated the ecological validity by
35

comparing users’ perception and behavior in immersive virtual environment (IVE) and real
environment (RE).
Some studies investigated if users’ environment appraisals and space perception in VE are
analogous to RE experience (Figure 2-6) (De Kort et al. 2003; Westerdahl et al. 2006; Kuliga et al.
2015). They all revealed the vast potential of VR as an empirical research tool to be applied in HBI.
The issues of experience in IVE were associated with lack of haptics feedback and navigation (De
Kort et al. 2003; Paljic 2017; Kuliga et al. 2015), atmospherics (missing sensory input, absence of
people, illumination/colors) (Kuliga et al. 2015), detailing (Kuliga et al. 2015), perception of
distances (Kuliga et al. 2015), and absence of sounds (De Kort et al. 2003; Westerdahl et al. 2006;
Kuliga et al. 2015). Studies all agreed that improving any of these issues would contribute to a
thorough sense of presence in IVE. Moreover, a few studies examined the use of VR in daylight and
lighting research (Chen, Cui, and Hao 2019; Kynthia Chamilothori, Wienold, and Andersen 2019).
Several conclusions can be drawn from two studies: VR can present lighting attributes of
open/close, diffuse/glaring, bright/dim, which are compatible with the physical environment (Chen,
Cui, and Hao 2019; Kynthia Chamilothori, Wienold, and Andersen 2019). VR lighting environment
was rated as an environment that is closer to the physical environment than video and photo
reproductions (Chen, Cui, and Hao 2019). There were no significant differences in perceptual
evaluations, reported physical symptoms in RE and IVE, and users reported a high level of presence
(Figure 2-7) (Kynthia Chamilothori, Wienold, and Andersen 2019). The same questions arise from
these studies: to what extent does the IVE have to resemble RE in order to trigger the analogous
effects on users? Moreover, is a fully experienced sense of presence necessary in all kinds of studies?
36

Figure 2-6 Exploring user experience RE (left) VE (right) (Kuliga et al. 2015)

Figure 2-7 VR for perception of daylit RE (left) VE (right) (Kynthia Chamilothori, Wienold, and Andersen 2019)

2.4 Summary
This chapter concluded seven design features from the literature review, the physiological
background of measuring experience in terms of biometric data, and ecological validity of virtual
reality. Specifically, this chapter first generalized how each design feature can influence the working
environment. Next, it introduced the autonomic nervous system's role in recognizing emotions. It
demonstrated definitions of heart rate, heart rate variability, skin temperature, electrodermal
activities, and how they can assist in quantifying experience. In the discussion of virtual reality,
37

ecological validity was considered essential to this research's reliability and applicability. Presence
and immersion are important indicators of ecological validity. Additionally, studies of the presence
of VR concluded that VR could be applied as experiment apparatus in certain studies and improving
immersion factors will enhance the sense of presence.

38

Chapter 3 Methodology
The proposed platform that integrates virtual reality, machine learning, and biosensors tends
to quantify the human perception of design features. The essential architectural design features and
their variables are identified through literature reviews. These design features are combined to
generate seven scenes of office models, and each scene includes one design feature with two
variables.
30 USC students have attended this study. They were asked to fill out a demographic survey
prior to the experiment. While wearing devices to collect biometric data, participants evaluated one
variable of a design feature each time and answered the questionnaire. Four types of biosignals
EDA, skin temperature, heart rate (HR), and heart rate variability (HRV), were collected by two
devices (Empatica Embrace watch and Garmin vivosmart 3) (Figure 3-1).
At the end of the experiment, synchronized data combined with subjective appraisals and
continuous data of biosignals were collected. Different algorithms (logistic regression, random
forest, support vector classifier, artificial neural network) are used to train the model, and the
accuracy score is compared to select the best fit model.
The research was reviewed and approved by the Institutional Review Board (IRB) reviewers
at the University of Southern California.

39

Figure 3-1 Methodology Diagram
3.1 Experiment Preparation
The design features and their variables were first determined in the experiment preparation
phase. The office model built in Revit was imported into a high-definition render pipeline in Unity.
After importing furniture assets and setting lighting values, environmental parameters were tuned to
generate seven scenes with high rendering quality. The scenes were cast into Oculus Quest 2.
Participants were able to navigate through rooms, interact with doors and answer questionnaires in
virtual office. Empatica Embrace and Garmin Vivosmart 3 were adopted to measure heart rate,
heart rate variability, EDA, skin temperature, respectively. In addition, participants were asked to
answer two questionnaires before and after the experiment.
40

3.1.1 Identify Key Architectural Design Features and Their Variables
The design features are categorized into three categories, including daylight& artificial
lighting, biophilia& view, and look& feel. In particular, design features applied in this study include
lighting source, illuminance level, plants, green view, color (ceiling, wall, floor), texture, and ceiling
height. The design features and variables are selected based on the literature review (Table 3-1).
These design features have been validated that they have considerable impacts on occupants’ health,
well-being, and productivity. This study does not include the interactive effects of multiple design
features. Only one feature will be examined each time. While participants review variables of one
feature, variables of other features remain default.
Table 3-1 Design features and variables
Design Features Default Variable A Variable B
light source (curtain) daylight artificial daylight
illuminance level (daylight +
artificial lighting)
normal low high
plants no no yes
green view no no yes
color (ceiling, wall, floor) dark palette red palette white palette
texture concrete fabric wood
ceiling height 3.2m 2.6m 3.8m

3.1.2 Rendering workflow of office model in Unity
The office model and feature variations were generated in Revit (Figure 3-2). The model
then was imported into High-Definition Render Pipeline (HDRP) in Unity. The HDRP is a high-
fidelity Scriptable Render Pipeline that highlights physically based lighting techniques, linear lighting,
41

HDR lighting, and a configurable hybrid tile/cluster deferred/forward lighting architecture, allowing
users to create photorealistic rendering (Unity Technologies n.d.).

Figure 3-2 Floor plan of the office in Revit
The investigator designed a small modern office with assets that were bought from Unity
Technologies Japan and Unimodels. After importing the 3D models, lighting sources were added to
the scene, including directional light as a daylight source, area lights, and spotlights for interior
lighting. Unity needs each light source's intensity, direction, distance, and color (Figure 3-3). Unity
provides global illumination, which is a system that combines direct and indirect lighting to provide
realistic lighting simulation. The virtual office adopted a baked global illumination to reduce CPU
and GPU costs during runtime. Moreover, there are no moving objects in the scene that require
real-time lighting. A baked global illumination generates a lightmap, a texture overlayed on objects
that stores pre-calculated lighting results such as brightness and color (Unity Technologies n.d.)
(Figure 3-4). This texture figure stores information of shadow and lighting, and it is one of ten
lightmaps (along with other ten direction lightmaps) of the office with artificial lighting.
42

Figure 3-3 Area light and light settings

Figure 3-4 Based lightmap of one room
43

The HDRP uses a volume framework (Unity Technologies n.d.). A volume contains
properties that can change environmental settings such as shadow, fog, indirect lighting, exposure,
sky, etc. This virtual office employed a global volume and two local volumes for each office room.
Global Volumes affect the camera wherever the camera is. On the contrary, Local Volumes affect
the view when the camera is in the Volume (Figure 3-5). The Global Volume was applied to define
overall environmental settings such as Tonemapping, Shadows, Ambient Occlusion (ambient light
that was cast by details), and so forth. The Local Volume controls the individual room based on the
experiment's setting. For instance, in the window view group, two offices have different High-
dynamic-range imaging (HDRI) sky settings. HDRI sky uses a cube map texture to represent the sky
(Figure 3-6). It also provides indirect lighting generated by the sky. The property values are
interpolated between Global Volumes and Local Volumes based on weight and priority.
44

Figure 3-5 Global Volume settings

Figure 3-6 HDRI of landscape view, n.d., PolyHaven, (Majboroda n.d.)
The materials of objects were adjusted to deliver a realistic rendering result (e.g., shader, UV
map tiling scale, metallic, smoothness, normal map) (Figure 3-7). Emissive materials were added to
45

the light source, which created an effect of bounced light combining with a light probe or bloom
property in volume (Figure 3-8). In the lighting scene, the investigator adjusted lightmapping settings
such as direct samples (number of samples used for direct lighting), indirect samples, environmental
samples, light bounces, filtering (reduce noise), and lightmap resolution before baking the lightmaps
(Figure 3-9). Next, a baked reflection probe was used to capture the scene and store the data in a
texture that can be overlayed on reflective objects such as glass, windows, and computer screens
(Figure 3-10). Prefabs (create reusable objects in each scene with the same parent), lighting settings,
Volume profiles were used across different scenes to facilitate the workflow (Figure 3-11).

Figure 3-7 Material setting of glass
46

Figure 3-8 emissive material and bloom, no bloom effect (left), bloom effect (right).

Figure 3-9 lightmap setting
47

Figure 3-10 Baked reflection probe cast reflection on reflective surface; no reflection probe (upper one)

Figure 3-11 Office Model in Unity

48

3.1.3 Generate VR office scenes
As mentioned before, the degree of presence is an increasing function of immersion factors.
Therefore, HDRP was selected to enhance the sense of presence through advanced lighting
simulation, photorealistic textures, and high-quality rendering. Additional office furniture assets
(bought from Unimodels and Unity Technology Japan) have been added to the scene. Further,
OpenXR and Oculus Quest 2 jointly add VR attributes (e.g., interaction, stereoscopy, and
navigation) to Unity scenes (Figure 3-12). OpenXR is an open API standard that allows high-
performance access to Extended Reality. Oculus Quest 2 is an all-in-one VR headset developed by
Meta (Table 3-2). Currently, Unity HDRP (version 2021.1.5) only supports VR in Oculus Rift; thus,
the program cannot be run independently on Oculus Quest 2. The Quest 2 headset was connected
to a laptop through a cable during the experiment.
The virtual office contains an XR rig, UI instruction, a survey system, and scene
management.
The virtual office used the XR interaction toolkit from Unity. It is a component-based
interaction system that creates VR or AR experience from Unity input events. It provides cross-
platform XR controller input, basic interaction, ray interactor, haptic feedback (vibration of the
controller), VR camera, and other extended functionalities (Unity Technologies n.d.). The XR rig
includes a camera to represent users’ views, a left-hand and a right-hand controller with hand
models, and a locomotion system (Figure 3-13). The left-hand controller provides continuous
movement via joystick, survey activation on the button “X,” and scene loader on button “Y.” Users
can answer the questionnaire using the ‘trigger’ button on the left-hand controller. The right-hand
controller only provides teleport movement and snap turn (Figure 3-14). To add customized
49

functions to the system, the investigator first bound input action with path (e.g., button, joystick),
then subscribe and unsubscribe from events in scripts (Figure 3-15).
Table 3-2 Oculus Quest 2 Hardware Specifications (Oculus 2020)
Device Oculus Quest 2
Screen Type LCD
Resolution per eye 1832 x 1920
Screen Refresh Rate 72Hz at launch, 90Hz to come
Weight 503 grams
Tracking Internal cameras
Processor Qualcomm Snapdragon XR2
RAM 6GB
Battery Life 2-3 hours

Figure 3-12 Workflow of generating VR scenes
50

Figure 3-13 XR rig and UI interactor on the left-hand controller

Figure 3-14 Teleport move
51

Figure 3-15 Subscribe and unsubscribe method ActivateUIMode
UI interaction was designed to give instructions to participants during the experiment. The
investigator believed that verbal instruction would break the sense of immersion. UI instruction
displays messages based on the users’ location, how long they have been at the location, and whether
they have finished the survey. The position of UI instruction is constantly updated based on the
position of the users’ camera in the Update method (the update method is called every frame).
The in-VR questionnaire contains users’ satisfaction, motivation, stress, and concentration
assessments in every room. The questionnaire UI is activated based on the controller’s position
when the function is called. When “submit” is clicked, the program creates a JSON file and stores it
locally. The JSON file contains the survey’s title, four answers, and the time they entered the room,
which is the event onset in later analysis. The time of entering the room was stored when users
stepped into the trigger area of each room (Figure 3-18).
The scene loader can be activated by the button “Y.” It reduces computational cost since the
computer only needs to render two rooms each time.
The investigator added physics interaction to the scene to make the scene more realistic. All
the walls, columns, and furniture have colliders, so the user would not pass through the models
52

(Figure 3-16). The doors with hinge joints were set on the same layer with the hand collider, and
thus users can interact with the door to create a more realistic experience (Figure-3-17).

Figure 3-16 Colliders of furniture in the scene

Figure 3-17 Participant uses hand to push a door
53

Figure 3-18 How the enter time is logged. (The cylinder represents the user. The console logs the time when the user entered Trigger
Area.)
Instead of navigating through the scenes physically in this experiment, participants navigated
the scenes via controllers. In addition to the typical continuous navigation mode, teleporting mode
was also provided to participants. Participants were allowed to choose one of the navigation modes.
The reasons are many VR users reported motion sickness when using continuous movement mode
via controllers; second, the biometric data will be significantly affected if participants physically
navigate through the scene, thus, impairing the accuracy of differentiation of emotional states. The
interaction and capability of navigation in IVE were identified as essential factors of the level of
presence. Along with high rendering quality and advanced lighting simulation, this study would be
able to deliver VR scenes with a higher level of presence than previous studies. Consequently, an
enhanced sense of presence will improve the reliability of the study.
Finally, this program applied several techniques to optimize performance. Occlusion culling
only renders objects within the camera's view, reducing rendering calculation cost. Level of details is
54

a technique that renders objects close to the camera at full fidelity. In addition, batching allows Unity
to combine objects and render them together.
3.1.4 Measure of Physiological Responses and Subjective Assessment
This study adopted a combination of subjective assessment and objective assessment. The
objective assessment (physiological data) was measured by two wearable watches, Empatica
Embrace and Garmin Vivosmart 3 (Figure 3-19). The Empatica Embrace watch measured skin
temperature and electrodermal activity, whereas the heart and heart rate variability were measured by
Garmin Vivosmart 3. The interval of each raw data of heart rate and heart rate variability is one
minute. The data can be exported after the measurement. However, the data exported from Garmin
Vivosmart 3 is not in CSV format. A Python program was applied to convert the file into a CSV file.
Figure 3-19 shows the specification of these two wearable watches. Note that Garmin converted
heart rate variability to stress level, ranging from 0 to 100 (0-25: resting state; 26-50 low stress; 51-75:
medium stress; 76-100 high stress) (GARMIN n.d.). How Garmin converted the data is not
accessible, and Garmin claimed that HRV data is not available for analysis by end consumers using
Garmin products (GARMIN n.d.). Therefore, the investigator uses heart rate and stress level for
analysis.
55

Figure 3-19 Specification of Biometric Sensors

The subjective evaluation consists of three parts. The demographic survey was conducted
before experiments (Table 3-3). The purpose of the demographic survey is to collect demographic
information from participants, and importantly, whether they had VR experience. This study
assumes that participants who had VR experience before might produce biometric data differently
than participants who used the VR headset for the first time.
In the second part, the in-game 5-scale semantic differential questionnaire was designed to
evaluate each feature (Table 3-4). The questionnaire includes the overall satisfaction level of the
office, the level of motivation to work, concentration, and stress level. Participants took the in-game
survey while evaluating the variables of each virtual office. The number of questions was limited to
four questions since participants had to answer fourteen times through the experiment.
56

Finally, participants were required to report a perceived sense of presence through three
questions (Table 3-5). This questionnaire is one of the measurements to assess the level of presence.
It has been adopted by many other studies (Sanchez-Vives and Slater 2005; Westerdahl et al. 2006).
One open-ended question intends to get feedback from participants regarding how to improve the
sense of immersion. Final question asks whether they had motion sickness during the experiment.
Table 3-3 Demographic survey
Demographic survey
Participant ID:
Date:
Age:
Gender:
Ever had VR experience (more than 2
hours): Yes/ No

Table 3-4 Questionnaire: Participants’ evaluation of each feature
-2 -1 0 1 2
Are you satisfied with the office design?
Do you feel motivated to work in this environment?
Do you feel any stress in this environment?
Do you think you can concentrate well in this environment?

Points scale: Strongly negative: -2; Negative: -1; Neutral 0; Positive: 1; Positive: 2

57

Table 3-5 Final questionnaire
-2 -1 0 1 2

To what extent did you have a
sense of being in the office?
Not at all ⃝ ⃝ ⃝ ⃝ ⃝ Very much so
How much time did you feel
the office is ’‘real' and forget
the ‘real world’?
Never ⃝ ⃝ ⃝ ⃝ ⃝

Almost all the time
When you think back on your
experience, do you think of
the office as images you saw,
or more as somewhere you
visit?
Only as images ⃝ ⃝ ⃝ ⃝ ⃝ Somewhere I visited

3.1.5 Data Synchronization
It is essential to synchronize physiological data and the corresponding stimulus. Common
time metrics were used to match the data from different devices. When participants save the survey
in VR, one JSON object is created. This JSON object contains the survey title, the survey answers,
and the time that participants entered the virtual room were stored in JSON format (Figure 3-20.
Later the time that participants entered the room was set as event onset.

Figure 3-20 Example of a JSON object
58

3.2 Experiment Procedure
The experiment was held at the 312A office of Watt Hall at University of Southern
California during weekdays (Figure 3-21). The 32 participants are students at USC.

Figure 3-21 The investigator tests the program in the lab (wearing mask and face cover)
In the beginning, participants were asked to wear biosensors and fill out a demographic
survey. After the demographic survey, the investigator explained research objectives and instructions
to participants and taught participants how to use the controllers to interact with the scene. The
investigator notified participants they could stop the experiment at any time if they felt dizzy and
uncomfortable. Next, participants put on the VR headset and helped open the cast function in
Quest 2, thus they could share the view with the investigator. Participants then entered the
preparation scene in VR and waited for the investigator’s instruction. The preparation stage helps
collect the baseline of physiological data and rule out the possibilities that the physiological
responses gleaned from participants are due to their first-time experience of VR. The investigator
did not talk to participants during the experiment unless they had questions.
59

In the experiment phase, participants experienced one preparation room and 14 offices
(Figure 3-25) (See Appendix. D). The participants started in the hallway to choose any one of two
rooms to enter. Participants learned the design feature in the hallway in case participants forgot to
evaluate the particular feature when they entered the office (Figure 3-23). Participants entered the
room by using their virtual hands to push the door. They were allowed to use continuous move,
teleport move, or a mix of both methods to navigate through the scene. They assessed the office
environment and answered the questionnaire by following UI instructions (Figure 3-24). Participants
loaded to the next scene after finishing one group of offices. After completing all 14 offices (Figure
3-22) (See Appendix. A), participants answered the questionnaire of presence to report their sense of
presence and opinions on how to improve the sense of immersion. Finally, participants acquired
their compensation and left the real room.
60

61

Figure 3-22 Design features and variables of 14 offices
62

Figure 3-23 UI instruction and feature reminder

Figure 3-24 Questionnaire in VR
63

Figure 3-25 Experiment Procedure.
3.3 The COVID-19 Guidelines
This experiment followed in-person social and behavioral research instruction under the
COVID-19 circumstance.
All participants have agreed to experiment and confirmed their health conditions.
Participants who entered the campus were required to receive vaccines and COVID-19 test within
one week. Both investigator and participant wore the mask during the whole experiment.
Participants were required to sanitize their hands after entered the lab. The lab maintained a high
ventilation rate, and participants were required to wear disposable face covers while wearing the VR
device (Figure 3-26). Between experiments, the VR face cover, controllers, and biosensors were
sanitized by alcohol wipes.
64

Figure 3-26 Disposable VR face cover
3.4 Data analysis
The data analysis includes four parts. First part is examining whether subjects exhibit the
same preference over a specific variable in a virtual office environment among the architectural
design features concluded from the literature review. A Shapiro-wilk test was conducted to test the
normality of the data in each group. Interval plots and Wilcoxon signed-rank tests were applied to
analyze the questionnaire in Minitab and SPSS. Minitab is a statistics package developed at the
Pennsylvania State University by researchers Barbara F. Ryan, Thomas A. Ryan, Jr., and Brian L.
Joiner in conjunction with Triola Statistics Company in 1972. SPSS is a statistical software suite
developed by IBM.
The second part investigates if the differences of biometric data between two design
variables of each feature are statistically significant. Paired t-test was adopted to compare heart rate,
stress level, and skin temperature.
65

The biometric data was synchronized with the event onset. The details are clarified in
chapter 4. The EDA data was imported into Ledalab v3.4.9. Ledalab provides analysis of EDA data
in MATLAB. The EDA was decomposed into skin conductance level (SCL) and skin conductance
response (SCR) by Continuous Decomposition Analysis (Benedek and Kaernbach 2010). Figure 3-
27 illustrates the analysis interface of Ledalab. The slowly varying wave refers to SCL activity (gray
wave), whereas the rapid changing wave represents SCR activity (blue wave). The red vertical line
represents the events (15 events include baseline and 14 variables). The intervals of events are not
consistent because the participant took different time to evaluate each room. The response window
was defined as 1 to 5 seconds. The minimum amplitude threshold was set to 0.01 μS. The event-
related analysis includes variables shown in Table 3-1. Event-related data was plotted per participant
(Figure 3-28). The sign test was used to examine the differences of sum of amplitudes of SCR
(AmpSum), area of phasic driver within time window (ISCR), and global mean value (Table 3-6).
Compared to traditional through to peak method, EDA scores analyzed by the decomposition
method were reported to be more sensitive (Benedek and Kaernbach 2010). The bar chart of mean
value of biometric data and event-related response plots (-1s to 5s) of phasic data were applied to
visualize the difference.
66

Figure 3-27 EDA analysis in Ledalab

67

Table 3-6 Event-related Analysis Variables (Benedek and Kaernbach 2010)

Variable
(Labels used in exported
files)
Event data
Event.nr Sequence number of event/marker
Event.nid Numerical ID of event
event.name Optional name or decription of event
Event.ud Optional userdata associated with event
CDA.nSCR
Number of significant (= above-threshold) SCRs within response
window (wrw)
CDA.Latency Response latency of first significant SCR wrw [s]
CDA.AmpSum
Sum of SCR-amplitudes of significant SCRs wrw (reconvolved
from corresponding phasic driver-peaks) [muS]
CDA.SCR
Average phasic driver wrw. This score represents phasic activity
wrw most accurately, but does not fall back on classic SCR
amplitudes [muS]
CDA.ISCR
Area (i.e. time integral) of phasic driver wrw. It equals SCR
multiplied by size of response window [muS*s]
CDA.PhasicMax Maximum value of phasic activity wrw [muS]
CDA.Tonic Mean tonic activity wrw (of decomposed tonic component)
TTP.nSCR
Number of significant (= above-threshold) SCRs within response
window (wrw)
TTP.AmpSum Sum of SCR-amplitudes of significant SCRs wrw [muS]
TTP.Latency Response latency of first significant SCR wrw [s]
Global Measures
Global.Mean Mean SC value within response window (wrw)
Global.MaxDeflection Maximum positive deflection wrw
Description
Continuous Decomposition Analysis (CDA)
(Extraction of Continuous Phasic/Tonic Activity based on Standard Deconvolution)
Standard trough-to-peak (TTP) or min-max analysis
68

Figure 3-28 Event-related plot of subject 06 (time window -1 to 5 seconds; features: plants and ceiling height)
The third part tests the ability of machine learning model to predict participants’ preferences
(Figure 3-29). The most common supervised learning tasks are regression (predicting value) and
classification (predicting classes). Multiclass classifiers predict five scales of each question:
satisfaction, motivation, stress, and concentration. A binary classifier predicts preferences that were
extracted from the questionnaire.
Irrelevant and redundant features were removed from the data. Instead of heart rate, stress
level, and skin temperature, the delta values of the data between baseline and each variable were
used to train the model. Missing values were handled. Ordinal Encoder from scikit-learn library
transferred target value into ordinal data. One-hot Encoder transferred categorical data (event name)
into dummy variables. Dummy variables use only ‘0’ and ‘1’ to represent a categorical value.
The final data contains 419 rows of data, where each row represents one design variable.
There are 13 features containing information on biometric data. Five groups of targets were
69

extracted from the data: satisfaction, motivation, stress, concentration, and preference. The input X
and output data y were split into training data and test data (80% training data). The test data was
used to examine the model’s performance on unseen data.
Logistic Regression (LR), random forest (RF), support vector classifier (SVC), and artificial
neural network (ANN) were selected as estimators. Fine-tuning the model is essential to the
performance of the model (Jin, Wang, and Sun 2021). Unsuitable hyperparameters will lead to a
failed model that overfits or underfits data (Figure 3-30). The best model exhibits the best
performance and generalizability on the unseen dataset. A standard approach to tweak the model
automatically is a grid-search combined with k-fold cross-validation. Grid search experiments with
every combination of input hyperparameters and try to find the optimum values of hyperparameters.
The k-fold cross-validation of one dataset is implemented k times, and each k sets is used as test set
once (Figure 3-31) (Jin, Wang, and Sun 2021).
The machine learning models were trained in Google Collaboratory. Google Collaboratory
(Colab) is a product from Google Research that is suited to machine learning in Python language.
Artificial Neural Network (ANN) implementation was based on the TensorFlow library. Other
machine learning techniques were based on scikit-learn Python library.
During the training, the model minimizes the loss between predicted value ŷ and target value
y. Accuracy is a commonly employed metric to evaluate the performance of the classifier, and
accuracy may not be a good measure when a false negative is unwanted (Müller, A., & Guido 2018).
Therefore, a comprehensive metrics, confusion matrix, was employed to evaluate the classification.
The confusion matrix is a two-by-two array, where the rows represent true classes, and the columns
denote predicted classes (Figure 3-32). Accuracy, precision, recall, and f-score that are derived from
70

the confusion matrix were used to compare models’ performance (Equation 3-1, 3-2, 3-3, 3-4). ROC
curve was adopted to visualize the performance comparison.

Figure 3-29 machine learning workflow and future works
71

Figure 3-30 Trade-off between bias and variance. (VanderPlas 2016)

Figure 3-31 K-fold cross validation (scikitlearn.org)
72

Figure 3-32 confusion matrix (Müller, A., & Guido 2018)
Equation 3-1 Accuracy
𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 =
𝑇𝑃 + 𝑇𝑁 𝑇𝑃 + 𝑇𝑁 + 𝐹𝑃 + 𝐹𝑁
Equation 3-2 Precision
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 =
𝑇𝑃 𝑇𝑃 + 𝐹𝑃
Equation 3-3 Recall
𝑅𝑒𝑐𝑎𝑙𝑙 =
𝑇𝑃 𝑇𝑃 + 𝐹𝑁
Equation 3-4 F1 score
𝐹 = 2 ∗
𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 ∗ 𝑟𝑒𝑐𝑎𝑙𝑙 𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 + 𝑟𝑒𝑐𝑎𝑙𝑙
The last part summarized participants’ sense of presence during the experience. The open-
ended questions provided instructions to improve the immersive experience.
3.5 Summary
This chapter described experiment preparation, experiment procedure, and data analysis. In
the experiment preparation phase, a virtual office with seven design features was made via Revit and
Unity. Three questionnaires and biosensors collected the subjective assessments and objective data.
73

The biometric data was synchronized with the questionnaires. The Wilcoxon signed-rank test was
adopted to compare participants’ evaluations in two design variables. The paired t-test was applied
to compare participants’ heart rate, stress level, and skin temperature. In the Ledalab software, the
original EDA data was decomposed to SCR and SCL. The sign test was employed to compare the
differences of sum of amplitude, area of phasic driver, and global mean value. The third part
investigated the machine learning model’s performance of predicting participants’ preferences. The
final part examined participants’ sense of presence reported after the experiment.

74

Chapter 4 Data sample and data synchronization
This chapter contains the description of the data sample, raw data, and how the datasets
were preprocessed for analysis. The data collected in the experiment includes two categories:
questionnaire and biometric data. The questionnaire category consists of a demographic
questionnaire, an in-VR questionnaire of office assessments, and a questionnaire of presence.
Biometric data contains heart rate, stress level, electrodermal activity, and skin temperature of each
subject during the experiment. All data was converted and documented in CSV Format.
The experiment was conducted at the 312A office, Watt Hall, University of Southern
California. The experiment date ranges from 1/22/2022 to 2/9/2022. 32 USC students have
participated in this experiment, and 30 valid datasets were collected. The majority of the participants
are graduate students at the School of Architecture, and others are undergraduate students and
Ph.D. students from different majors.
4.1 Data Sample and Raw Data of Questionnaires
The questionnaire category contains a demographic questionnaire, an in-VR questionnaire of
office assessments and a questionnaire of the sense of presence. The investigator gave experiment
instruction in English or Chinese. All participants have responded to the questionnaires in English,
and they were capable of understanding the questionnaire thoroughly.
4.1.1 Demographic questionnaire
Participants answered the demographic questionnaire before the experiment. This
questionnaire includes participants’ ID, date of experiment, gender, age, and whether they had VR
75

experience for more than two hours. To maintain privacy and confidentiality, the participants are
identified through participant ID. Two data sets were removed because participants did not finish
the experiment due to severe motion sickness. No missing data and abnormal data were identified in
this questionnaire.
30 questionnaires were documented for the analysis (Table 4-1). The experiment date ranges
from 1/22/2022 to 2/9/2022. Participants were in the age range between 20 to 35 (mean: 26.2,
STDEV: 3.83), including 17 females and 13 males. 24 participants had not experienced VR for more
than two hours.
Table 4-1 Demographic questionnaire
Demographic Questionnaire
num Participant ID (date +
sequence)
Date Age Gender Ever had VR experience (more
than 2 hours): Yes/ No
1
SA1
22-Jan 28 female no
2
SB1
23-Jan 24 male yes
3
SB2
23-Jan 24 male no
4
SB3
23-Jan 23 male no
5
SC1
24-Jan 31 female yes
6
SD1
25-Jan 26 male no
7
SE1
26-Jan 33 male no
8
SE2
26-Jan 23 female no
9
SE3
26-Jan 26 female no
10
SE4
26-Jan 34 male no
11
SE5
26-Jan 25 female no
12
SE6
26-Jan 23 female no
13
SF1
31-Jan 35 male no
14
SF2
31-Jan 31 female no
15
SF3
31-Jan 22 male yes
16
SF4
31-Jan 24 female no
17
SF5
31-Jan 24 male no
18
SG1
1-Feb 28 female no
19
SH1
2-Feb 24 male no
20
SH2
2-Feb 25 female no
21
SH3
2-Feb 28 female yes
22
SH4
2-Feb 26 male yes
76

23
SH5
2-Feb 26 female no
24
SI1
2-Feb 24 female no
25
SI2
2-Feb 20 female no
26
SI3
2-Feb 30 female no
27
SJ1
4-Feb 25 male yes
28
SK1
9-Feb 22 female no
29
SK2
9-Feb 22 female no
30
SK3
9-Feb 30 male no

4.1.2 In-VR Questionnaire of Office Assessments
Participants answered this questionnaire in the VR environment. The data includes
participants’ ID, participants’ assessments for 14 offices with different design features (Table 4-2).
Participants went through design features in order of “lighting source,” “illuminance level,”
“window view,” “indoor plants,” “color,” “texture,” and “ceiling height.” However, variables were
chosen by participants randomly. For instance, in a lighting source scene, participants knew variables
of daylight and artificial lighting in advance. The investigator set the variables randomly (the left
office was not always set as positive variables).
Participants evaluated the office environment in terms of satisfaction, motivation to work,
stress, and concentration levels in 5 differential scales (-2, -1, 0, 1, 2). Note that participants
answered stress levels on a scale that “2” represents “very stressful” (Table 4-3 shows the original
scale). In order to have a better illustration of the data, the stress scale was later multiplied by -1.
Missing data were identified in the questionnaire. Participants might forget to choose an
answer or forget to click the “submit” button.
Appendix. B demonstrates 30 sets of modified data. The answers for stress level were
multiplied by -1. For instance, stress level of subject 3 changed from “-1, -1, -1, 0, -2 …” to “1, 1, 1,
0, 2 …”.
77

Table 4-2 Data sample of the in-VR questionnaire
num ID
Event.Name satisfaction motivation stress concentrate
1 SA1
baseline
1 SA1
arti lighting -1 -1 0 1
1 SA1
daylight 1 -1 0 2
1 SA1
highIllu 0 -2 -1 -1
1 SA1
lowIllu 1 0 1 1
1 SA1
cityView -1 -1 -1 -1
1 SA1
greenView 1 -1 2 2
1 SA1
Plants 1 0 1 1
1 SA1
noPlants -1 -2 0 -1
1 SA1
bright -1 -1 0 -1
1 SA1
red -2 -1 -2 -1
1 SA1
fabric -2 -1 -1 -1
1 SA1
wood -1 -1 -1 -2
1 SA1
highCeiling 1 -1 0 1
1 SA1
lowCeiling 0 0 -1 1
2 SB1
baseline
2 SB1
arti lighting 1 0 -1 1
2 SB1
daylight 2 0 1 1
2 SB1
highIllu 2 1 2 -1
2 SB1
lowIllu 1 0 0 1
2 SB1
cityView 2 1 2 1
2 SB1
greenView 2 1 2 0
2 SB1
Plants 2 1 2 1
2 SB1
noPlants 1 0 2 0
2 SB1
bright 0 -1 1 -2
2 SB1
red 1 1 1 1
2 SB1
fabric 1 0 0 0
2 SB1
wood 1 1 2 -1
2 SB1
highCeiling 1 1 1 1
2 SB1
lowCeiling -2 -2 -2 -2

78

4.1.3 Questionnaire of Presence
This questionnaire is one of the measurements to assess the level of presence (Deniaud et al.
2015)(Table 4-3). It has been adopted by many previous studies (Sanchez-Vives and Slater 2005;
Westerdahl et al. 2006). One open-ended question intends to get feedback from participants
regarding how to improve the sense of immersion. The final question asks whether they had
uncomfortable motion sickness during the experiment.
The first three questions investigate the sense of immersion. The open-ended question
provided examples such as adding sound, avatar, NPC (human), more interaction with the scene,
better rendering, wireless device. Participants were encouraged to give feedback rather than these
examples. Participants were asked to report motion sickness when their feelings reached an
uncomfortable level. 30 questionnaires were listed in Appendix C.
Table 4-3 Data sample of questionnaire of presence
num ID To what
extent did
you have a
sense of
being in the
office? (Not
at all -2, -1, 0,
1, 2 very
much so)
How much
time did you
feel the office is
'real', and you
forgot the 'real
world'? (Never -
2, -1, 0, 1, 2
almost all the
time)
When you
think back of
your
experience,
do you think
the office as
images that
you saw, or
more as
somewhere
that you visit?
(Only as
images -2, -1,
0, 1, 2
somewhere I
visited)
Adding what factor
can improve your
sense of immersion?
(e.g., sound, avatar
(have a body), more
interaction with the
scene, better
rendering, add NPC,
wireless etc.?)
Did you
have
motion
sickness?
1 sa1 0 -1 2 sound, better
rendering
no
2 sb1 0 -1 1 NPC, sound no
3 sb2 1 -1 0 sound yes

79

4.2 Data Sample and Raw Data of Biometric data
The heart rate and heart rate variability were measured by Garmin Vivosmart 3. The interval
of each raw data of heart rate and heart rate variability is one minute. The heart rate ranges from 0 to
200 BMP. The Garmin Vivosmart converted heart rate variability to the stress level, ranging from 0
to 100 (0-25: resting state; 26-50 low stress; 51-75: medium stress; 76-100 high stress). The Empatica
Embrace sensor measured skin temperature and electrodermal activity (EDA). The biometric data
from two devices were synchronized with the corresponding stimulus.
4.2.1 Heart Rate and Stress Level
Heart rate and stress level can be exported from Garmin by date. A Python program
converted the data from FIT format to CSV format. The data sample demonstrates the heart rate
and stress level of participant 06 on 1/25/2022 and how the data was preprocessed (Table 4-7). The
raw data includes a timestamp of heart rate with an interval of one minute, heart rate value,
timestamp of stress level, and stress level value (Table 4-4). Event id and event time were logged by
the VR office program (Table 4-5). The biometric data and event time was synchronized based on
time metrics. Each event’s heart rate and stress level were rounded (Table 4-6). For example, the
event time of baseline is 14:22:02 (second is less or equal to 30), then the heart rate for baseline
adopted the original heart rate at 14:22:00. The event time of high illuminance level was 14:27:53
(second is larger than 30), and the heart rate for high illuminance level used the original heart rate at
14:28:00.
Note that the timestamp of heart rate is not continuous. According to Garmin, the device
may lose track of heart rate due to the connection issue between skin and devices. For instance,
compressing the area underneath the watch can result in losing data (GARMIN n.d.). Moreover, the
80

stress level could be 0 or -1, and the reason is not explained by Garmin. Stress level data with -2, -1,
0, 1 were removed before analysis.
Table 4-4 Original heartrate and stress level of subject 06
time (date) timestamp_16 heartrate Stress level time Stress level value
1/25/22 14:20 14:20:00 70 14:22:00 11
1/25/22 14:23 14:23:00 69 14:23:00 1
1/25/22 14:25 14:25:00 70 14:24:00 0
1/25/22 14:26 14:26:00 69 14:25:00 0
1/25/22 14:27 14:27:00 71 14:26:00 0
1/25/22 14:28 14:28:00 69 14:27:00 1
1/25/22 14:30 14:30:00 72 14:28:00 0
1/25/22 14:31 14:31:00 71 14:29:00 0
1/25/22 14:32 14:32:00 69 14:30:00 0
1/25/22 14:33 14:33:00 67 14:31:00 1
1/25/22 14:34 14:34:00 69 14:32:00 0
1/25/22 14:35 14:35:00 71 14:33:00 0
1/25/22 14:36 14:36:00 73 14:34:00 0
1/25/22 14:37 14:37:00 72 14:35:00 0
1/25/22 14:40 14:40:00 70 14:36:00 0
1/25/22 14:41 14:41:00 69 14:37:00 0
1/25/22 14:43 14:43:00 71 14:38:00 1
1/25/22 14:44 14:44:00 86 14:39:00 0
1/25/22 14:45 14:45:00 82 14:40:00 1
14:41:00 1
14:42:00 0
14:43:00 -2

Table 4-5 Event time of subject 06
title enterTime
baseline 14:22:02
daylight 14:23:02
arti lighting 14:24:42
lowIllu 14:26:30
highIllu 14:27:53
greenView 14:29:25
cityView 14:30:35
noPlants 14:31:52
81

Plants 14:33:15
bright 14:34:41
red 14:36:10
fabric 14:37:22
wood 14:38:45
lowCeiling 14:39:55
highCeiling 14:41:05

Table 4-6 Rounded heartrate for each event (subject 06)
timestamp_16 heartrate event time hr round
14:20:00 70 14:22:02 70
14:23:00 69 14:23:02 69
14:25:00 70 14:24:42 70
14:26:00 69 14:26:30 69
14:27:00 71 14:27:53 69
14:28:00 69 14:29:25 72
14:30:00 72 14:30:35 71
14:31:00 71 14:31:52 69
14:32:00 69 14:33:15 67
14:33:00 67 14:34:41 71
14:34:00 69 14:36:10 73
14:35:00 71 14:37:22 72
14:36:00 73 14:38:45

14:37:00 72 14:39:55 69
14:40:00 70 14:41:05 69
14:41:00 69

14:43:00 71

14:44:00 86

14:45:00 82

82

Table 4-7 raw data and data synchronization of subject 06

4.2.2 Electrodermal activity
The Empatica Embrace watch measured electrodermal activity (EDA). The data can be
exported by date. The EDA data contains the timestamp expressed as Unix Time in milliseconds
(UTC) and EDA data in microSiemens(µS) with a sampling rate of 4Hz (column 1 and column 2 in
Table 4-11). Unix Time is the number of seconds elapsed since January 1, 1970 (UTC).
Column 3 converts Unix Time (UTC) to Pacific Standard Time through a formula “(Unix
Time/86400/1000) + DATE (1970,1,1) -(8/24)” (Table 4-8). For subject 06, 14:22:02 is the event
time for baseline (Table 4-8). 14:21:00 (time of baseline minus one minute) was set as start time for
analysis (column 4 in Table 4-8). The time increases with an increment of 0.25, representing a
sampling rate of 4 Hz. Event ID in column 5 represents the order of offices that participants
entered. Participants spent different lengths of time answering the questionnaire and learning to
navigate through controllers, and they were able to enter rooms randomly in one design feature
time (date) timestamp_16 heart_rateevent time hr round stress_level_time stress_level_value event time stress round evet id title enterTime
1/25/22 14:20 14:20:00 70 14:22:02 70 14:22:00 11 14:22:02 11 1 baseline 14:22:02
1/25/22 14:23 14:23:00 69 14:23:02 69 14:23:00 1 14:23:02 1 2 daylight 14:23:02
1/25/22 14:25 14:25:00 70 14:24:42 70 14:24:00 0 14:24:42 0 3 arti lighting 14:24:42
1/25/22 14:26 14:26:00 69 14:26:30 69 14:25:00 0 14:26:30 0 4 lowIllu 14:26:30
1/25/22 14:27 14:27:00 71 14:27:53 69 14:26:00 0 14:27:53 0 5 highIllu 14:27:53
1/25/22 14:28 14:28:00 69 14:29:25 72 14:27:00 1 14:29:25 1 6 greenView 14:29:25
1/25/22 14:30 14:30:00 72 14:30:35 71 14:28:00 0 14:30:35 0 7 cityView 14:30:35
1/25/22 14:31 14:31:00 71 14:31:52 69 14:29:00 0 14:31:52 0 8 noPlants 14:31:52
1/25/22 14:32 14:32:00 69 14:33:15 67 14:30:00 0 14:33:15 0 9 Plants 14:33:15
1/25/22 14:33 14:33:00 67 14:34:41 71 14:31:00 1 14:34:41 1 10 bright 14:34:41
1/25/22 14:34 14:34:00 69 14:36:10 73 14:32:00 0 14:36:10 0 11 red 14:36:10
1/25/22 14:35 14:35:00 71 14:37:22 72 14:33:00 0 14:37:22 0 12 fabric 14:37:22
1/25/22 14:36 14:36:00 73 14:38:45 14:34:00 0 14:38:45 0 13 wood 14:38:45
1/25/22 14:37 14:37:00 72 14:39:55 69 14:35:00 0 14:39:55 0 14 lowCeiling 14:39:55
1/25/22 14:40 14:40:00 70 14:41:05 69 14:36:00 0 14:41:05 0 15 highCeiling 14:41:05
1/25/22 14:41 14:41:00 69 14:37:00 0
1/25/22 14:43 14:43:00 71 14:38:00 1
1/25/22 14:44 14:44:00 86 14:39:00 0
1/25/22 14:45 14:45:00 82 14:40:00 1
14:41:00 1
14:42:00 0
14:43:00 -2
subject 06 hr and stress
83

scene. As a result, different subjects have different duration of analysis time and different event
sequences.
Some participants exhibit EDA data below 0.1 µS. According to Empatica, a small portion
of the general population does not show significant EDA changes in an experimental setting. Other
factors such as temperature, humidity, and medication also impact EDA data. The sensor exhibits
negative EDA values because of accuracy and tolerance. Based on production tests, the absolute
accuracy at 0 point is (-0.5 µS, +0.5 µS), and 98.3% of the samples are in the (-0.01 µS, +0.01 µS)
range (Empatica n.d.).
Table 4-9 lists EDA data of subject 06 in the format required by the software Ledalab. It
includes time for analysis, EDA data, event ID, and a placeholder.
Table 4-8 EDA data sample and data preprocess of subject 06
Unix
Timestamp
(UTC)
EDA (microS) convert time time for analysis Eid
1.64E+12 0.607421 14:21:00 0.00 0
1.64E+12 0.609222 14:21:00 0.25 0
1.64E+12 0.612199 14:21:00 0.50 0
1.64E+12 0.611916 14:21:00 0.75 0
1.64E+12 0.614729 14:21:01 1.00 0
1.64E+12 0.617827 14:21:01 1.25 0
1.64E+12 0.619152 14:21:01 1.50 0
1.64E+12 0.618264 14:21:01 1.75 0
1.64E+12 0.620397 14:21:02 2.00 0
1.64E+12 0.614373 14:21:02 2.25 0
1.64E+12 0.610357 14:21:02 2.50 0
1.64E+12 0.612623 14:21:02 2.75 0
1.64E+12 0.614099 14:21:03 3.00 0
… … … … …
1.64E+12 0.734128 14:22:00 60.50 0
1.64E+12 0.735383 14:22:00 60.75 0
1.64E+12 0.734053 14:22:01 61.00 0
84

1.64E+12 0.732803 14:22:01 61.25 0
1.64E+12 0.731986 14:22:01 61.50 0
1.64E+12 0.732697 14:22:01 61.75 0
1.64E+12 0.732319 14:22:02 62.00 1
1.64E+12 0.731147 14:22:02 62.25 0
1.64E+12 0.728942 14:22:02 62.50 0
1.64E+12 0.72629 14:22:02 62.75 0
1.64E+12 0.72247 14:22:03 63.00 0
1.64E+12 0.7213 14:22:03 63.25 0
1.64E+12 0.718254 14:22:03 63.50 0
1.64E+12 0.716948 14:22:03 63.75 0
… … … … …
1.64E+12 0.813761 14:23:01 121.00 0
1.64E+12 0.814209 14:23:01 121.25 0
1.64E+12 0.816277 14:23:01 121.50 0
1.64E+12 0.820312 14:23:01 121.75 0
1.64E+12 0.823602 14:23:02 122.00 2
1.64E+12 0.830195 14:23:02 122.25 0
1.64E+12 0.830648 14:23:02 122.50 0
1.64E+12 0.84061 14:23:02 122.75 0
… … … … …
1.64E+12 1.064333 14:24:41 221.00 0
1.64E+12 1.065476 14:24:41 221.25 0
1.64E+12 1.065794 14:24:42 221.50 3
1.64E+12 1.064281 14:24:42 221.75 0
1.64E+12 1.065875 14:24:42 222.00 0
… … … … …
1.64E+12 1.151086 14:26:28 328.00 0
1.64E+12 1.135507 14:26:28 328.25 0
1.64E+12 1.129562 14:26:29 328.50 0
1.64E+12 1.151027 14:26:29 328.75 0
1.64E+12 1.154547 14:26:29 329.00 0
1.64E+12 1.154555 14:26:29 329.25 0
1.64E+12 1.155234 14:26:30 329.50 4
1.64E+12 1.154485 14:26:30 329.75 0
1.64E+12 1.154406 14:26:30 330.00 0
1.64E+12 1.171285 14:26:30 330.25 0
85

… … … … …
1.64E+12 2.598882 14:41:04 1204.25 0
1.64E+12 2.567659 14:41:04 1204.50 0
1.64E+12 2.557945 14:41:04 1204.75 0
1.64E+12 2.551709 14:41:05 1205.00 15
1.64E+12 2.555184 14:41:05 1205.25 0
1.64E+12 2.581734 14:41:05 1205.50 0
1.64E+12 2.642547 14:41:05 1205.75 0
1.64E+12 2.702329 14:41:06 1206.00 0
… … … … …
1.64E+12 2.720262 14:42:08 1268.25 0
1.64E+12 2.715511 14:42:08 1268.50 0
1.64E+12 2.710912 14:42:08 1268.75 0
1.64E+12 2.706356 14:42:09 1269.00 0
1.64E+12 2.703926 14:42:09 1269.25 0
1.64E+12 2.703314 14:42:09 1269.50 0
1.64E+12 2.710189 14:42:09 1269.75 0
1.64E+12 2.728878 14:42:10 1270.00 0
1.64E+12 2.748574 14:42:10 1270.25 0
1.64E+12 2.76046 14:42:10 1270.50 0
1.64E+12 2.766089 14:42:10 1270.75 0

Table 4-9 EDA in the format for Ledalab (subject 06)
0.00 0.607421 0 0
0.25 0.609222 0

0.50 0.612199 0

0.75 0.611916 0

1.00 0.614729 0

1.25 0.617827 0

1.50 0.619152 0

1.75 0.618264 0

2.00 0.620397 0

2.25 0.614373 0

2.50 0.610357 0

2.75 0.612623 0

3.00 0.614099 0

86

3.25 0.616715 0

3.50 0.617681 0

3.75 0.618548 0

4.00 0.619915 0

… …
…
61.25 0.732803
0

61.50 0.731986
0

61.75 0.732697
0

62.00 0.732319
1

62.25 0.731147
0

62.50 0.728942
0

62.75 0.72629
0

63.00 0.72247
0

… …
…
328.50 1.129562
0

328.75 1.151027
0

329.00 1.154547
0

329.25 1.154555
0

329.50 1.155234
4

329.75 1.154485
0

330.00 1.154406
0

330.25 1.171285
0

330.50 1.169305
0

330.75 1.183168
0

331.00 1.202229
0

331.25 1.213172
0

… …
…
1204.00 2.607104
0

1204.25 2.598882
0

1204.50 2.567659
0

1204.75 2.557945
0

1205.00 2.551709
15

1205.25 2.555184
0

1205.50 2.581734
0

1205.75 2.642547
0

1206.00 2.702329
0

1206.25 2.634475
0

87

1206.50 2.584958
0

1206.75 2.588599
0

1207.00 2.631968
0

… …
…

4.2.3 Skin Temperature
The Empatica Embrace measures skin temperature in degrees on the Celsius scale every
second. The data sample exhibits Unix Time (UTC) and skin temperature data in degrees, and the
Unix Time was converted to Pacific Standard Time (Table 4-10).
The skin temperature was synchronized with the event time. The average skin temperature
for each event is an average of degrees within one minute after event onset. No missing data was
detected. Table 4-11 lists mean skin temperature of subject 06 for each event.
Table 4-10 Data sample of skin temperature
Unix
Timestamp
(UTC)
Degrees
(Â°C)
convert
time
1.64E+12 32.111 14:20:00
1.64E+12 32.111 14:20:01
1.64E+12 32.137 14:20:02
1.64E+12 32.137 14:20:03
1.64E+12 32.137 14:20:04
1.64E+12 32.137 14:20:05
1.64E+12 32.137 14:20:06
1.64E+12 32.137 14:20:07
1.64E+12 32.137 14:20:08
1.64E+12 32.163 14:20:09
1.64E+12 32.137 14:20:10
1.64E+12 32.137 14:20:11
1.64E+12 32.163 14:20:12
1.64E+12 32.163 14:20:13
1.64E+12 32.163 14:20:14
1.64E+12 32.163 14:20:15
88

1.64E+12 32.163 14:20:16
1.64E+12 32.163 14:20:17
… … …

Table 4-11 Mean skin temperature of each event (subject 06)
Eid Event name
Temp mean
1 Baseline
32.5702
2 Daylight
32.60102
3 Artificial lighting
32.67741
4 Low illuminance
32.90221
5 High illuminance
32.98831
6 Green view
33.02961
7 City view
33.07725
8 No plants
33.16043
9 Plants
33.26193
10 Bright
33.26918
11 Red
33.28069
12 Fabric
33.27131
13 Wood
33.42523
14 Low ceiling
33.51954
15 Heigh ceiling
33.61715

4.3 Data for machine learning
The integrated data contains event-related analysis exported from Ledalab, heart rate, stress
level, skin temperature, and answers to the in-VR questionnaire. The delta values of heart rate, stress
level, skin temperature between baseline and each variable per participant were calculated and added
to the data.
Machine learning models cannot handle missing values. Missing value in the feature
‘CDA.latency,’ ‘TTP.latency,’ and questionnaires were filled in with 0. In event-related analysis,
latency refers to the time latency of the first significant SCR within the time window. Latency was
empty when none of the significant SCR was detected within the time window. In the questionnaire,
89

0 scale indicates a neutral state. Missing values of heart rate and stress level were filled with the mean
value.
Feature engineering involves removing irrelevant and redundant features in the data and
combining existing features to a more useful one (feature extraction). Feature ‘Num,’ ‘ID,’ and
‘Event.Nr’ were removed since they are irrelevant to analysis results. Feature ‘heart rate,’ ‘stress
level,’ and ‘skin temperature’ were dropped from the data because the delta values would be used to
train the model. Finally, the ‘baseline’ rows were removed since it was used to glean the baseline of
physiological data. The answers of satisfaction, motivation, stress, and concentration were combined
into a new feature: ‘Preference.’ The value is ‘like’ when the average answer is larger than 0
otherwise, the value is ‘dislike.’ The Pearson Correlation between numerical values were conducted
to examine the data redundancy (Figure 4-1). A highly strong correlation exists between
“CDA.AmpSum,” “SCR,” and “ISCR” and between “CDA.tonic” and “global.mean”. AmpSum,
SCR, ISCR were all calculated based on skin conductance responses within the time window. The
tonic value determines EDA data's tendency, thus strongly correlated to the global mean value. The
“ISCR” and “CDA.tonic” data were preserved for training.
90

Figure 4-1 The heatmap of Pearson correlation of numerical values
The final data contains 419 rows of data, where each row represents one design variable.
There are 13 features containing information on biometric data (Table 4-12) (See Appendix. E). Five
groups of targets were extracted from the data: satisfaction, motivation, stress, concentration, and
preference.
Before preparing training and test data, Ordinal Encoder from scikit-learn library transferred
target value into ordinal data. One-hot Encoder transferred categorical data (event name) into
dummy variables. Dummy variables use only ‘0’ and ‘1’ to represent a categorical value. The input
data X was transformed into the shape (n_samples, n_features), whereas the target data y is in the
shape of (n_samples). The input and output data were split into training data and test data (80%
training data). The test data was used to examine the model’s performance on unseen data.

91

num ID Event.Nr Event.Name CDA.nSCR CDA.Latency CDA.AmpSum
1 SA1 1 baseline 0

0
1 SA1 2 arti lighting 6 1 0.096448
1 SA1 3 daylight 0

0
1 SA1 4 highIllu 1 4.25 0.012478
1 SA1 5 lowIllu 0

0

CDA.SCR CDA.ISCR CDA.PhasicMax CDA.Tonic TTP.nSCR TTP.Latency
0.006449 0.10318645 0.126894 0.357776 0

0.025268 0.40428079 0.151989 0.384785 0

0.001366 0.02185025 0.056798 0.51392 1 5
0.007333 0.11732606 0.1151 0.528872 1 4
0.006203 0.09925081 0.136959 0.533671 0

TTP.AmpSum Global.Mean Global.MaxDeflection HR hr-sub stress
level
0 0.399609 0.005661 77 0 30
0 0.461636 0.006662 74 -3 24
0.025301 0.52386 0.006369 77 0 32
0.015912 0.543737 0.013909 76 -1

0 0.546228 0.012128 78 1 36

stress-
sub
temp temp-
sub
satisfaction motivation stress concentrate preference
0 27.76136 0

-6 27.77505 0.013689 -1 -1 0 1 dislike
2 27.71556 -0.0458 1 -1 0 2 like

27.6589 -0.10246 0 -2 -1 -1 dislike
6 27.65315 -0.10821 1 0 1 1 like
Table 4-12 Top 5 rows of the integrated data

4.4 Summary
This chapter contains questionnaire data and biometric data during the experiment.
Questionnaire data consists of a demographic questionnaire, an in-VR questionnaire, and a
questionnaire of presence. The experiment date ranges from 1/22/2022 to 2/9/2022. Participants
were in the age range between 20 to 35 (mean: 26.2, STDEV: 3.83), including 17 females and 13
92

males. 24 participants had not experienced VR for more than two hours. Biometric data includes
heart rate and stress level, which were collected by Garmin Vivosmart 3. Empatica Embrace 2 was
used to measure electrodermal activity and skin temperature. The data sample, data preprocessing,
and raw data were explained. The ISCR value across all participants range from 0 to 5.384 µS·s. The
heart rate ranges from 51 to 117 (bpm), the stress level ranges from 0 to 93, and the skin
temperature ranges from 24.966 to 34.243 (°C). There are 74 missing values in heart rate and 72
missing values in stress level. The final data shape for machine learning is (419, 26). 11 participants
(37%) have reported uncomfortable motion sickness. 2 participants stopped the experiment, and
their data was removed from the dataset.

93

Chapter 5 Results and discussion
This chapter includes the analysis of the in-VR questionnaire, the analysis of biometric data,
testing the ability of machine learning models to predict participants’ preferences, and the summary
of the questionnaire of presence. After examining the normality and skewness of the data in each
group, A Wilcoxon signed-rank test was performed to compare participants’ subjective evaluations
in each group of design variables. The paired t-test was adopted to investigate the differences in
heart rate, stress level, and skin temperature in each design feature. A sign test was used to compare
the sum of amplitude, SCR, and ISCR. The significance level of p-value was marked (*: marginally
significant, **: significant, ***: very significant). The investigator employed logistic regression,
support vector classifier, random forest, and artificial neural network to predict the preferences. All
models adopted grid search and cross-validation techniques to search for the best hyperparameters.
The analysis of the questionnaire of presence reveals the immersion levels reported by participants
and feedbacks on improving the immersive experience.
5.1 Analysis of the in-VR questionnaire
A Shapiro-Wilk test reveals that the questionnaire data in each group is not normally
distributed. For instance, the Shapiro-Wilk p-values of satisfaction, motivation, stress, and
concentration of lighting source are p daylightSatis=.001, p daylightMotiv=.002, p daylightStr=.006, p daylightCon=.000,
p artificialSais=.003, p artificialMotiv=.045, p artificialStr=.001, p artificialCon=.027, which are all less than p=.05.
Therefore, the investigator conducted Wilcoxon signed-rank tests to compare participants’
evaluations in seven pairs of design variables and interval plots to visualize the differences.
94

The interval plot provides a comparison of the variance of participants’ evaluation with two
design variables (Figure 5-1). Overall, participants gave a higher rate to the office with daylight. An
overlap can be found in the assessment of concentration levels. There are statistical differences in
satisfaction, motivation, and stress between the two variables.

Figure 5-1 Interval plot of lighting source

Wilcoxon signed-rank tests were performed to compare participants’ evaluations of
satisfaction level, motivation to work, stress level, and concentration in virtual offices with daylight
and artificial lighting (Table 5-1). Significant differences can be observed in all four types of
assessments. Participants reported higher satisfaction levels in the daylit office (M=1.20, SD=0.71)
than levels in the office with artificial lighting (M=0.5, SD=1.11), Z=-3.143, p=.002. Participants
95

stated they felt more motivated to work in the office with daylight (M=1.17, SD=0.87) than in the
office with artificial lighting (M=0.00, SD=1.31), Z=-3.69, p=0.001. Participants reported that they
felt more stressed in the office with artificial lighting (M=0.07, SD=0.89) than levels in the office
daylight (M=0.97, SD=0.99), Z=-3.20, p=.001. Finally, participants expressed that they could
concentrate better if they would work in the office with daylight (M=1.07, SD=0.87) than in the
office with artificial lighting (M=0.37, SD=1.19), Z=-2.80, p=.005.
This result correlates with the outcome from the literature review that occupants prefer
daylight over artificial lighting (Boubekri 2008; Ander 2003; Galasiu and Veitch 2006). Although
participants only perceive the lighting source visually in VR headset, their evaluations of satisfaction,
motivation, stress, and concentration with daylight are statistically higher than evaluations of the
office with artificial lighting. Therefore, in this virtual working environment setting, daylight
positively impacts participants’ satisfaction, motivation, stress, and concentration.
Table 5-1 Wilcoxon signed-rank test of lighting source
Lighting source
Satisfaction Motivation Stress Concentration
P (two-tailed) 0.007*** 0.001*** 0.002*** 0.007***

Participants gave a slightly higher average rate to the office with a high illuminance level
(Figure 5-2). Overlaps can be found in all four evaluations, and the large interval indicates a large
amount of variation of each design variable.
96

Figure 5-2 Interval plot of illuminance level
Based on the Wilcoxon signed-rank test, a moderately significant difference can only be
observed in the assessment of satisfaction level (Table 5-2). Participants reported higher satisfaction
levels in the office with high illuminance level (M=0.93, SD=0.91) than in the office with low
illuminance level (M=0.3, SD=1.09), Z=-1.986, p=.047.
Overall, participants prefer a high illuminance level to a low illuminance level; however, one
cannot draw the conclusion that it is strongly associated with the results from literature reviews.
Despite the lighting setting being physically based in the HDRP of Unity, lighting can be affected by
a series of rendering parameters such as exposure, indirect lighting intensity, and color tone. A
method that can measure the illuminance level more accurately is needed in order to verify that
occupants prefer illuminance levels higher than levels recommended by professional organizations
(Boubekri 2008; Logadóttir, Christoffersen, and Fotios 2011).
97

Table 5-2 Wilcoxon signed-rank test of Illuminance level
Illuminance level
Satisfaction Motivation Stress Concentration
P (two-tailed) 0.047** 0.111 0.582 0.212

Based on the interval plot, participants’ evaluation of satisfaction, motivation, and stress in
the office with a green view is significantly higher than those without a green view (Figure 5-3). An
overlap of the confidence interval can be found in the evaluation of concentration.

Figure 5-3 Interval plot of window view
Based on the Wilcoxon signed-rank test, significant differences can be observed in all
evaluations (Table 5-3). Participants reported higher satisfaction levels in the office with a green
view (M=1.77, SD=0.50) than those without a green view (M=0.73, SD=0.74), Z=-4.424, p=.000.
Participants expressed that they have more motivation to work in the office with a green view
(M=1.43, SD=0.97) than without a green view (M=0.43, SD=0.935), Z=-3.537, p=.000. Participants
98

felt less stressed to work in the office with a green view (M=1.43, SD=0.97) than in the office
without a green view (M=0.37, SD=0.89), Z=-3.686, p=.000. Participants reported higher
concentration levels in the office with a green view (M=1.167, SD=0.95) than those without a green
view (M=0.63, SD=0.81), Z =-2.173, p=.030.
According to the interval plot, participants expressed higher rates to the office with indoor
plants, and no overlaps are found between the two variables (Figure 5-4).

Figure 5-4 Interval plot of indoor plants
The Wilcoxon signed-rank test indicates that significant differences exist in all evaluations
(Table 5-5). Participants were more satisfied with the office with indoor plants (M=1.20, SD=0.61)
than the office without indoor plants (M=0.33, SD=0.71), Z=-3.965, p=.000. Participants felt more
motivated to work in the office with plants (M=1.23, SD=0.63) than without plants (M=0.20,
SD=0.961), Z=-4.099, p=.000. Participants reported lower stress levels in the office with plants
99

(M=1.0 SD=0.96) than without plants (M=0.07, SD=0.91), Z=-4.028, p=.000. Finally, the results
show that participants felt more concentrated in the office with plants (M=0.97, SD=0.81) than
without plants (M=0.30, SD=0.88), Z=-3.070, p=.002.
Participants’ evaluations of green window view and indoor plants are significantly higher
than the office without green view and plants. The results echoed the findings in the literature
review that natural environments appeal to occupants and incorporating plants into the office
promotes workplace satisfaction (Gray and Birrell 2014; I. Elzeyadi 2011; Africa and Sachs 2016).
Table 5-3 Wilcoxon signed-rank test of window view
Window view
Satisfaction Motivation Stress Concentration
P (two-tailed) 0.000*** 0.000*** 0.000*** 0.030**

Table 5-4 Wilcoxon signed-rank test of indoor plants
Indoor plants
Satisfaction Motivation Stress Concentration
P (two-tailed) 0.000*** 0.000*** 0.000*** 0.002***

The interval plot exhibits participants’ evaluations of satisfaction, motivation, and stress in
the office in beige color is significantly higher than those in red (Figure 5-5). There is no overlap of
the confidence interval found in the plot.
100

Figure 5-5 Interval plot of color
Based on the Wilcoxon signed-rank test, significant differences can be disclosed in all
evaluations (Table 5-5). Participants gave significantly higher rates of satisfaction (M=0.57, 1.01),
motivation to work (M=0.50, 1.08), stress level (M=0.07, SD=1.11), and concentration level
(M=0.40, SD=1.276) to the office in beige color. Their evaluations of the red office were
considerably lower, Z satisfaction=-3.999, p satisfaction=.000, Z motivation=-3.954, p motivation=.000, Z stress=-2.575,
p stress=.010, Z concentration=-3.723, p concentration=.000.
Only two colors were selected among the colors examined in the literature review, and the
results correspond with the literature review. Color affects occupants’ moods and performance
(Nancy Kwallek, Lewis, and Robbins 1988; Nancy Kwallek 1996; Mahnke 1996). In particular,
occupants mostly liked the beige color and least liked the red color since red office induces anxiety
and stress (Nancy Kwallek, Lewis, and Robbins 1988; Nancy Kwallek 1996; Mahnke 1996).
101

Table 5-5 Wilcoxon signed-rank test of color
Color
Satisfaction Motivation Stress Concentration
P (two-tailed) 0.000*** 0.000*** 0.010** 0.000***

The interval plot exhibits that participants’ evaluations of satisfaction and motivation in the
office with the wood surface are slightly higher than those with the fabric surface (Figure 5-6). On
the contrary, the plot illustrates that the evaluations of stress and concentration are marginally higher
in the office with fabric textures. Evidently, the plot shows wide confidence intervals indicating
considerable variation in each group.

Figure 5-6 Interval plot of texture
In the Wilcoxon signed-rank test, no significant difference was found in the data (Table 5-6).
(Z satisfaction=-0.316, p satisfaction=.752, Z motivation=-0.637, p motivation=.524, Z stress=-0.727, p stress=.467,
Z concentration=-0.698, p concentration=.485).
102

Studies that examined users’ texture preference in an office context were not found. The
large variation of participants’ evaluations indicates that participants gave similar rates to both
variables. During the experiment, some participants disliked texture due to the UV map scale, color
tone, or the product’s design (e.g., carpet pattern). Therefore, the texture feature needs to be
decomposed into more specific variables to be assessed, such as color and design.
Table 5-6 Wilcoxon signed-rank test of texture
Texture
Satisfaction Motivation Stress Concentration
P (two-tailed) 0.752 0.524 0.467 0.485

The interval plot reveals a statistical difference between evaluations of low ceiling height and
high ceiling height, and the amount of variation is comparable in each group (Figure 5-7).

Figure 5-7 Interval plot of ceiling height
103

Based on the Wilcoxon signed-rank test, significant differences can be found in all
evaluations (Table 5-7). Participants gave significantly higher rates of satisfaction (M=1.00, 0.80),
motivation to work (M=0.79, 1.01), stress level (M=0.69, SD=0.89), and concentration level
(M=0.83, SD=0.85) to the office with high ceiling height. Their evaluations of the office with low
ceiling height were noticeably lower, Z satisfaction=-4.050, p satisfaction=.000, Z motivation=-3.649, p motivation=.000,
Z stress=-3.749, p stress=.000, Z concentration=-3.808, p concentration=.000.
The results correlate with those in the literature review that occupants prefer a room with
higher ceiling height (Baird, Cassidy, and Kurr 1978; Vartanian et al. 2015).
Table 5-7 Wilcoxon signed-rank test of ceiling height
Ceiling Height
Satisfaction Motivation Stress Concentration
P (two-tailed) 0.000*** 0.000*** 0.000*** 0.000***

5.2 Analysis of biometric data
Participants’ heart rate, stress level, and skin temperature in the office with daylight and
office with artificial lighting were compared. On average, participants’ stress levels (M=43.00,
SD=30.43) and skin temperature (M=30.05, SD=1.96) are slightly higher in the office with daylight
than stress levels with artificial lighting (M stress=42.42, SD stress=30.87, M temperature=30.02,
SD temperature=1.96). Heart rate is lower in the office with daylight (M=83.3, SD=14.20) than with
artificial lighting (M=83.74, SD=16.05). The difference is not statistically significant (T hr=-0.797,
p hr=.433, T stress=0.197, p stress=.846, T temperature=0.540, p temperature=.594) (Table 5-8). Figure 5-8 illustrates
the phasic data of different subjects (1-8) within the response window. The comparison of mean is
104

shown in the figures (5-9, 5-10, 5-11, 5-12).

Figure 5-8 Event-related responses of lighting source
105

Figure 5-9 mean value of ISCR (lighting source)

Figure 5-10 mean value of heart rate (lighting source)
106

Figure 5-11 mean value of stress level (lighting source)

Figure 5-12 mean value of skin temperature (lighting source)
107

The sign test did not show statistically significant differences between the sum of amplitude
and global mean value in two offices, p AmpSum=.109, p mean=.855 (Table 5-9). However, the ISCR in
the office with daylight (M=0.036, SD=0.051) is significantly lower than the response in the office
with artificial lighting (M=0.142, SD=0.267), p=.006.
The analysis shows that participants were more aroused in the office with artificial lighting.
Aligned with the questionnaire analysis, it can be concluded that participants’ emotional intensity is
higher in the virtual office with artificial lighting. It is associated with negative emotions, including
stress, distraction, and less motivation. The change of heart rate, stress level, and skin temperature is
too minor to reveal any information.
Table 5-8 paired t-test of lighting source
Lighting source (paired t-test)
Heart rate Stress level Skin temperature
T-value -0.797 0.197 0.540
P-Value 0.433 0.846 0.594

Table 5-9 sign test of lighting source
Lighting source (sign test)
AmpSum ISCR Global mean
P-Value 0.109 0.006*** 0.855

Based on paired t-test, participants’ heart rate (M=86.04, SD=15.73), stress level (M=44.52,
SD=31.79), and skin temperature (M=30.38, SD=1.93) are slightly higher in the office with high
illuminance level than heart rate (M=84.00, SD=14.42), stress levels (M=40.92, SD=31.175), and
skin temperature (M=30.35, SD=2.03) in the office with artificial lighting. The difference is not
statistically significant (Thr=0.390, phr=.701, Tstress=1.038, pstress=.311, Ttemperature=0.661,
ptemperature=.514) (Table 5-10). Figure 5-13 illustrates the phasic data of different subjects within
108

the response window (others see appendix). The charts of the mean are shown in the figures (Figure
5-14, 5-15, 5-16, 5-17).
109

Figure 5-13 Event-related responses of illuminance level
110

Figure 5-14 mean value of ISCR (illuminance level)

Figure 5-15 mean value of heart rate (illuminance level)
111

Figure 5-16 mean value of skin temperature (illuminance level)

Figure 5-17 mean value of stress level (illuminance level)
112

The sign test did not show statistically significant differences between the sum of amplitude,
ISCR, and global mean value in two offices, p AmpSum=.109, p ISCR=.829, p mean=.855 (Table 5-9).
No statistically meaningful differences between physiological responses in the two offices
were observed. Some participants reported that they could not tell the differences between the two
rooms. The change of illuminance level is probably too minor to cause discrepancies.
Table 5-10 paired t-test of illuminance level
Illuminance level (paired t-test)
Heart rate Stress level Skin temperature
T-value 0.390 1.038 0.661
P-Value 0.701 0.311 0.514

Table 5-11 sign test of illuminance level
Illuminance level (sign test)
AmpSum ISCR Global mean
P-Value 1.000 0.822 0.361

On average, participants’ heart rate (M=85.39, SD=14.18) and skin temperature (M=30.65,
SD=2.03) are slightly higher in the office with a green view than heart rate (M=84.60, SD=12.81)
and skin temperature (M=30.61, SD=1.99) in the office without a green view. The stress level in the
green view office (M=43.65, SD=30.41) is lower than in the office without green view (44.36,
SD=30.87). The difference is not statistically significant (T hr=-0.637, p hr=.532, T stress=0.088,
p stress=.930, T temperature=-0.995, p temperature=.328) (Table 5-12). Figure 5-18 illustrates the phasic data of
different subjects within the response window (others see appendix). The charts of mean are shown
in the figures (Figure 5-19, 5-20, 5-21, 5-22).

113

Figure 5-18 Event-related responses of view
114

Figure 5-19 mean value of ISCR (view)

Figure 5-20 mean value of heart rate (view)
115

Figure 5-21 mean value of stress level (view)

Figure 5-22 mean value of skin temperature (view)
116

Statistically significant differences between the sum of amplitude, ISCR, and global mean
value in two offices were not found, p AmpSum=.308, p ISCR=.629, p mean=.111 (Table 5-13).
Despite considerable disparities in the result from the questionnaire, no statistically
important differences between biometric data in the two offices were observed. Participants were
more aroused in the office with a green view, and their stress levels were slightly lower in the office
with a green view.
Table 5-12 Paired t-test of view
window view (paired t-test)
Heart rate Stress level Skin temperature
T-value -0.637 0.088 -0.995
P-Value 0.532 0.930 0.328

Table 5-13 sign test of view
Window view (sign test)
AmpSum ISCR Global mean
P-Value 0.308 0.629 0.111

The paired t-test demonstrates that participants’ heart rate (M=86.25, SD=13.82) and stress
level (M=45.76, SD=31.26) are slightly higher in the office with plants than heart rate (M=84.41,
SD=14.60), stress levels (M=44.92, SD=31.73) in the office without plants. The difference is not
statistically significant (T hr=0.-1.759, p hr=.096, T stress=0.404, p stress=.690, T temperature=0.331,
p temperature=.743) (Table 5-14). Figure 5-23 illustrates the phasic data of different subjects within the
response window (others see appendix). The charts of the mean are shown in the figures (5-24, 5-25,
5-26, 5-27).
117

Figure 5-23 Event-related responses of indoor plants
118

Figure 5-24 mean value of ISCR (plants)

Figure 5-25 mean value of heart rate (plants)
119

Figure 5-26 mean value of stress level (plants)

Figure 5-27 mean value of skin temperature (plants)
120

The average ISCR in the office with plants (M=0.32, SD=1.01) is higher than ISCR in the
office without plants (M=0.18, SD=0.29). The sign test did not show statistically significant
differences between the sum of amplitude, ISCR, and global mean value in two offices,
p AmpSum=.600, p ISCR=.530, p mean=.629 (Table 5-15).
No statistically important differences between physiological responses in the two offices
were observed. The ISCR data reveals that participants were more aroused in the office with plants,
which can be linked to their positive responses to the design variable.
Table 5-14 Paired t-test of plants
plants (paired t-test)
Heart rate Stress level Skin temperature
T-value -1.759 0.404 0.331
P-Value 0.096* 0.690 0.743

Table 5-15 sign test of plants
plants (sign test)
AmpSum ISCR Global mean
P-Value 0.581 0.855 0.855

The paired t-test demonstrates that participants’ heart rate (M=86.73, SD=14.80) and stress
level (M=45.58, SD=30.82), and skin temperature (M=31.18, SD=1.93) are slightly higher in the red
office than heart rate (M=85.65, SD=16.35), stress levels (M=46.00, SD=32.01), and skin
temperature (M=31.16, SD=1.97) in the beige office. The difference is not statistically significant
(T hr=-1.712, p hr=.103, T stress=-0.391, p stress=.699, T temperature=-0.422, p temperature=.676) (Table 5-16).
Figure 5-28 illustrates the phasic data of different subjects within the response window (others see
appendix). The charts of the mean are shown in figures (Figure 5-29, 5-30, 5-31, 5-32).
121

Figure 5-28 Event-related responses of color
122

Figure 5-29 mean value of ISCR (color)

Figure 5-30 mean value of heart rate (color)
123

Figure 5-31 mean value of stress level (color)

Figure 5-32 mean value of stress level (color)
124

The average ISCR in the red office (M=0.30, SD=0.72) is higher than ISCR in the beige
office (M=0.19, SD=0.37). The sign test did not show statistically significant differences between
the sum of amplitude, ISCR, and global mean value in two offices, p AmpSum=1.000, p ISCR=.855,
p mean=.855 (Table 5-17).
No statistically important differences between physiological responses in the two offices
were observed. The ISCR data indicates that participants were more aroused in the red office.
Table 5-16 paired t-test of color
color (paired t-test)
Heart rate Stress level Skin temperature
T-value -1.712 -0.391 -0.422
P-Value 0.103 0.699 0.676

Table 5-17 sign test of color
color (sign test)
AmpSum ISCR Global mean
P-Value 1.000 0.855 0.855

The paired t-test demonstrates that participants’ heart rate (M=85.79, SD=16.31) in the
office with fabric texture is slightly higher than the date in the office with wood texture (M=83.88,
SD=14.29). In addition, the stress level in the office with wood texture (M=49.76, SD=30.00) is
slightly higher than stress levels (M=48.72, SD=29.70) in the office with fabric texture. The
difference is not statistically significant (T hr=1.206, p hr=.243, T stress=-0.976, p stress=.339, T temperature=-
1.344, p temperature=.190) (Table 5-18). Figure 5-33 illustrates the phasic data of different subjects within
the response window (others see appendix). The charts of the mean are shown in figures (Figure 5-
34, 5-35, 5-36, 5-37).
The average ISCR in the office with wood texture (M=0.35, SD=0.90) is higher than ISCR
in the office with a fabric texture (M=0.14, SD=0.26). The sign test did not show statistically
125

significant differences between the sum of amplitude, ISCR, and global mean value in two offices,
p AmpSum=.607, p ISCR=.855, p mean=.855 (Table 5-19).
No statistically important differences between physiological responses in the two offices
were identified. The ISCR data reveals that participants were more aroused in the office with wood
texture. It is not consistent with the questionnaire analysis since the two design variables did not
elicit substantial differences in participants’ assessments.
Table 5-18 paired t-test of texture
texture (paired t-test)
Heart rate Stress level Skin temperature
T-value 1.206 -0.976 -1.344
P-Value 0.243 0.339 0.190

Table 5-19 sign test of texture
texture (sign test)
AmpSum ISCR Global mean
P-Value 0.607 0.855 0.855

The two mean values of heart rate, stress level, and skin temperature are very close between
the office with low ceiling height and the office with high ceiling height. The difference is not
statistically significant (T hr=-0.236, p hr=.816, T stress=0.025, p stress=.980, T temperature=-1.085,
p temperature=.287) (Table 5-20). Figure 5-38 illustrates the phasic data of different subjects within the
response window (others see appendix). The charts of the mean are shown in figures (Figure 5-39,
5-40, 5-41, 5-42).
126

Figure 5-33 Event-related responses of texture
127

Figure 5-34 mean value of ISCR (texture)

Figure 5-35 mean value of heart rate (texture)
128

Figure 5-36 mean value of stress level (texture)

Figure 5-37 mean value of skin temperature (texture)
129

The average ISCR in the office with high ceiling height (M=0.24, SD=0.33) is slightly higher
than ISCR in the office with low ceiling height (M=0.23, SD=0.42). The sign test did not show
statistically significant differences between the sum of amplitude, ISCR, and global mean value in
two offices, p AmpSum=.332, p ISCR=.325, p mean=.198 (Table 5-21) (Figure-38, 5-39, 5-40, 5-41).
130

Figure 5-38 Event-related responses of ceiling height
131

Figure 5-39 mean value of ISCR (ceiling height)

Figure 5-40 mean value of heart rate (ceiling height)
132

Figure 5-41 mean value of stress level (ceiling height)

Figure 5-42 mean value of skin temperature (ceiling height)
133

No statistically important differences between physiological responses in the two offices
were identified. The ISCR data reveals that participants expressed the same level of arousals in two
offices, which is against the questionnaire analysis.
Table 5-20 paired t-test of ceiling height
Ceiling height (paired t-test)
Heart rate Stress level Skin temperature
T-value -0.236 0.25 -1.085
P-Value 0.816 0.980 0.287

Table 5-21 sign test of ceiling height
Ceiling height (sign test)
AmpSum ISCR Global mean
P-Value 0.332 0.325 0.198

5.3 Predicting participants’ preferences using machine learning models
All models were implemented in Google Colab using the Python language. The biometric
data heart rate, stress level, skin temperature was replaced by Δ heart rate, Δ stress level, Δ skin
temperature (compared to individual baseline). The missing value in “CDA.latency” and
questionnaire was filled with zero. The missing value in biometric data was imputed by the mean
value. The other strategy was dropping the rows that contain a missing value.
On the one hand, 113 rows of data would be dropped, compromising the performance. On
the other hand, there is only one value missing in the entire row in many cases, which might not
significantly affect the results. The Pearson correlation was conducted to remove redundant data. A
highly strong correlation exists between “CDA.AmpSum,” “SCR,” and “ISCR” and between
“CDA.tonic” and “global.mean”. AmpSum, SCR, ISCR were all calculated based on skin
conductance responses within the time window. The tonic value determines EDA data's tendency,
134

thus strongly correlated to the global mean value. The “ISCR” and “CDA.tonic” data were
preserved for training.
One set of data does not contain the evaluation of low ceiling, and it was removed from the
data. The final X data shape is (419, 26), and y data shape is (419,). Ordinal Encoder class
transferred target value to ordinal scale (dislike: 0, like: 1). One Hot Encoder class converted feature
‘event name’ to dummy variables. The data was split into the training set and test set at a ratio of
0.8:0.2. The class weights of target values were slightly biased (163 dislikes and 256 likes). Thus, a
stratification based on the y value was considered, and the resulting class weights of target data in
the test set were 33 (dislike) and 51 (like) (random state=8).
The grid search and cross-validation techniques were adopted to optimize the model by
fitting the training set. The test set was then fed to the model to validate the performance. Table 5-
22, 23, 24, 25 list the best hyperparameters of each model acquired through grid search. The
investigator increased the iteration times to converge the model (Figure 5-43).
Table 5-22 The best parameters of logistic regression
Logistic regression
Hyperparameters Best parameters
C 2.98
penalty l1

Table 5-23 The best parameters of SVC
SVC
Hyperparameters Best parameters
C 1000
gamma 0.001
kernel ‘rbf’

135

Table 5-24 The best parameters of random forest
Random forest
Hyperparameters Best parameters
criterion ‘gini’
max_depth 6
max_features ‘auto’
n_ estimators 100

Table 5-25 The best parameters of articial neural network
Artificial neural network
Hyperparameters Best parameters
n_neurons 300
n_hidden 2
learning rate 0.0038
activation function “elu”
kernel initializer “he_normal”

Figure 5-43 The converging process of the artificial neural network
136

The best model was utilized to predict ŷ value. The classification report of pairs (y - ŷ),
including precision score, recall score, f1-score, and accuracy score, was generated to evaluate the
performance of each model. The equations are listed in chapter 3. The classification reports generate
macro average (0.5 multiplied by the score of each class) and weighted average value simultaneously.
Due to the unbalanced weights, the weighted score was selected.
As listed in Table 5-26, the random forest model outperformed other models (precision
score: 0.73, recall score: 0.73, f1 score: 0.71, and accuracy score: 0.73). The ROC curve illustrates
that the random forest model demonstrated the best performance (Figure 5-44). The artificial neural
network has relatively low scores. Note that this report might not be the best scores. The data was
shuffled into the training set, and test set randomly, which may cause different training results.
Moreover, either a discrete parameter grid or an evenly distributed grid might miss the best
hyperparameters. The investigator has run the models many times. The scores were not consistent,
and the best model was not fixed. In binary classification, the true positive is C_ {1, 1} and the true
negative is C_ {0, 0}. A model with the highest recall score is preferred in a real-life scenario (less
false negatives). Therefore, the random forest model can be selected as the best model.
137

Figure 5-44 ROC curve of models
Table 5-26 The classification report in terms of weighted score
Model Precision recall f1-score accuracy
Logistic regression 0.70 0.70 0.69 0.70
SVC 0.68 0.65 0.66 0.65
Random forest 0.73 0.73 0.71 0.73
Artificial neural network 0.65 0.61 0.61 0.61

5.4 Analysis of the questionnaire of presence
The presence questionnaire intended to investigate participants’ sense of presence when they
experienced the VR office using a differential scale. According to the bar chart, this VR program
delivered an experience with a high sense of presence (Figure 5-45). The mean rating of question
one (M=1.3) and question three (M=1.3) are higher than the mean value of question two (M=0.6).
Participants reported that it was hard to forget the real world. The reasons can be summarized as the
heavy weight of the headset, the feeling of motion sickness, feelings of stuffy due to the mask and
138

face cover, the cable of the headset, hot weather. The results indicate the applicability of the results
to the real world.

Figure 5-45 The questionnaire of presence.q1: “to what extent did you have a sense of being in the office? (Not at all -2, -1, 0, 1, 2
very much so); q2: how much time did you feel the office is 'real,' and you forgot the 'real world'? (Never -2, -1, 0, 1, 2 almost all the
time); q3: when you think back on your experience, do you think of the office as images that you saw or more as somewhere that you
visit? (Only as images -2, -1, 0, 1, 2 somewhere I visited).”
Figure 5-46 illustrates participants’ opinions on improving the sense of immersion. Sound is
the most desired element to create an immersive virtual environment. Many participants proposed
that human characters should be added to the office, and the investigator was concerned that the
unrealistic non-player character (human models) would decrease the sense of immersion. Other
desired aspects include better rendering, more interaction with the scene, live objects, and so forth.
These aspects should be considered in future studies (Figure 5-47).
139

Figure 5-46 participants' feedbacks on improving the sense of presence
140

Figure 5-47 count of feedbacks from participants

5.5 Summary
The analysis of participants’ evaluations in light source, window view, indoor plants, colors,
and ceiling height in the virtual environment corresponds with the findings from the literature
review.
Daylight has positive impacts on participants’ satisfaction, motivation, stress, and
concentration. The natural environments appeal to occupants and incorporating plantings into the
office will boost the satisfaction level of the office. Among the two colors selected from the
literature review, red is the least favored color since it causes anxiety and stress as a surface color in
the office. In addition, occupants prefer a room with a higher ceiling height.
141

The comparisons of evaluations in illuminance level and texture are not strongly related to
the literature review. Overall, participants prefer the office with a high illuminance level, but the
differences in motivation, stress, and concentration are not statistically significant. Moreover, the
differences in evaluations of texture are not statistically significant.
The mean value of heart rate, stress level, skin temperature did not demonstrate significant
differences. The heart rate, stress level, and skin temperature that Empatica Embrace and Garmin
Vivosmart 3 measured during this experiment cannot reflect participants’ preferences. The skin
conductance responses show noticeable gaps between design variables. However, only ISCR values
between daylight and artificial lighting and heart rate values between plants and no plants are
statistically significant. The skin conductance response has the potential to represent participants’
emotional arousal in each office.
The machine learning models can predict users’ preferences at accuracy as high as 73% on
the validation dataset. The recall score is a more desired metric in a real-life scenario; therefore, the
random forest model outperformed other models with a weighted average recall score of 0.73.
The reports from participants indicate that this VR office program had delivered an
experience with a strong sense of presence. To improve the sense of presence, sound, non-player
character, better rendering, and more interaction with the scene are the most desired additional
aspects for an immersive virtual environment. Some physical limitations, including the heavyweight
of the headset, the feeling of motion sickness, stuffy feelings due to the mask and face cover, the
cable of the headset, hot weather, had compromised the immersive experience. Overall, the results
have verified the results' applicability to the real world.
In conclusion, researchers can adopt this virtual office to investigate participants’ evaluations
of design features such as lighting source, biophilia, colors, and ceiling height, and the results can be
142

applied to the real world. The correlation between users’ preferences and users’ biometric data was
not statistically verified. However, the machine learning model was able to predict users’ preferences
with an accuracy of 73%.

143

Chapter 6 Conclusion and future work
6.1 Conclusion
Architectural design features as one of the main components of an indoor environment are
features that configure an indoor space and can be visually identified. Understanding design features’
influence on humans can help designers build an indoor environment that reduces occupants’ stress
levels and improves well-being and productivity. However, there is a lack of empirical evidence of
the impacts of the interior environment on human experience and a method to quantify the impacts.
In addition, there is no applicable laboratory setting to investigate the impacts of design features.
Neuroscience provides tools to measure and quantify the impacts on occupants, and virtual reality
can be applied as experimental apparatus to study occupants’ responses. Machine learning
algorithms have the ability to predict user preferences based on design features and biometric data
inputs.
Seven design features and their variables were researched through a literature review: lighting
source, illuminance level, window view, indoor plants, color, texture, and ceiling height. Heart rate,
heart rate variability, skin temperature, electrodermal activities were selected as users’ physiological
metrics. Studies provided evidence that arousal is linked to an increase in electrodermal responses,
and valence is more associated with heart rate and heart rate acceleration and deceleration.
Ecological validity is crucial to the reliability and applicability of the research. It determines whether
the human experience and responses triggered in the virtual environment are similar to those
triggered in the real world.
144

Design variables were combined to generate seven scenes of office models in Revit and
Unity. 32 USC students participated in the study. They were asked to fill out a demographic survey
prior to the experiment. While wearing devices to collect biometric data, participants evaluated one
variable of a design feature each time and answered the questionnaire. Four types of biosignals:
EDA, skin temperature, heart rate (HR), and heart rate variability (HRV), were collected by two
devices (Empatica Embrace watch and Garmin vivosmart 3). After the experiment, a questionnaire
with ordinal scales was adopted to measure the sense of presence; it also collected participants’
feedback on improving the immersion and whether they had motion sickness.
All questionnaires and biometric data were converted and documented in CSV format. The
questionnaire and biometric data were synchronized based on event onset. EDA was decomposed
into skin conductance response in Ledalab. The event-related responses with a minimum amplitude
of 0.05 μS and a time window of 1- 5s were selected. Heart rate and stress level were rounded based
on event onset. The data includes the average skin temperature for each event within one minute
after event onset. The subjective and biometric data were integrated and preprocessed for machine
learning models.
After examining the normality and skewness of the data in each group, A Wilcoxon signed-
rank test was performed to compare participants’ subjective evaluations in each group of design
variables. The paired t-test was adopted to investigate the differences in heart rate, stress level, and
skin temperature in each design feature. A sign test was used to compare the sum of amplitude,
SCR, and ISCR. The investigator employed logistic regression, support vector classifier, random
forest, and artificial neural network to predict the preferences. All models adopted grid search and
cross-validation techniques to search for the best hyperparameters. The classification report of
prediction of validation set was compared to select the best model.
145

Daylight has positive impacts on participants’ satisfaction, motivation, stress, and
concentration. The natural environments appeal to occupants and incorporating plants into an office
will boost the satisfaction level of the occupants. Among the two colors selected from the literature
review, red is the least favored color since it causes anxiety and stress as a surface color in the office.
In addition, occupants prefer a room with a higher ceiling height. However, the results from the
illuminance level and texture analysis are not strongly related to the literature review (Table 6-1).
Table 6-1 Summary of subjective evaluation and biometric data
Design feature
Preferred design
variables
subject evaluations
that show statistically
significant differences
Whether
correlated to
literature
review
Biosignals that
show
statistically
significant
differences
lighting source daylight satisfaction, motivation,
stress, concentration
yes ISCR (very
significant)
illuminance level high illuminance
level
satisfaction partially no
window view green window view satisfaction, motivation,
stress, concentration
yes no
indoor plants incorporate plants satisfaction, motivation,
stress, concentration
yes heart rate
(marginally
significant)
color of surface beige satisfaction, motivation,
stress, concentration
yes no
texture of surface no preference none n/a no
ceiling height high ceiling height satisfaction, motivation,
stress, concentration
yes no

The heart rate, stress level, skin temperature between every pair of design variables did not
demonstrate statistically significant differences. A noticeable difference can be observed in each
design variable’s mean value of skin conductance responses. However, only ISCR values between
daylight and artificial lighting and heart rate between plants and no plants are statistically significant.
The machine learning models can predict users’ preferences at accuracy as high as 73% on the
146

validation dataset. The recall score is a more desired metric in a real-life scenario; therefore, the
random forest model outperformed other models with a weighted average recall score of 0.73.
The reports from participants indicate that this VR office program had delivered an
experience with a high sense of presence. To improve the sense of presence, sound, non-player
character, better rendering, and more interaction with the scene are the most desired additional
aspects for an immersive virtual environment. Some physical limitations, including the heavyweight
of the headset, the feeling of motion sickness, stuffy feelings due to the mask and face cover, the
cable of the headset, hot weather, had compromised the immersive experience. Overall, the results
have verified the applicability of the research to the real world.

6.2 Research limitation and future work
6.2.1 Better rendering and VR experience
Some participants reported that they would have a better experience using headset wirelessly
because they were trapped by the cable when they turned around physically. At this time, unity
HDRP is not compatible with Quest 2 (version 2021.1.5f); therefore, participants had to assess the
offices through Oculus rift. In the future, it is encouraged to run the program independently in the
headset to help better users forget reality.
Despite the applied optimization techniques, the high-resolution baked lightmaps took up
too much memory storage and significantly compromised the performance. Consequently, one
might experience drop frames, camera shaking, and poor rendering. The investigator had to reduce
the resolution of lightmaps from 3GB (for one scene) to 50MB (one scene). In addition, the
investigator decreased objects’ scales in lightmap to accelerate baking time, resulting in a less detailed
147

rendering of objects. Further studies are encouraged to run the program on a high-performance
computer and apply optimization techniques other than occlusion culling, level of details, and
batching to provide better rendering results and a smooth experience.
As identified in chapter 5, future studies are encouraged to incorporate these additional
features to virtual office such as sound, human characters, live objects, and more interaction with
the scene to improve the sense of presence.
6.2.2 Size, quality, and dimensions of the data
To further improve the prediction performance, the size and quality of the dataset have to
be improved. First, more participants with a mix of different ages, races, and professionals should be
recruited. Moreover, more advanced biosensors can be applied. For instance, despite the importance
of HRV to differentiation of human emotion, how Garmin converted the data is not accessible, and
Garmin claimed that HRV data is not available for analysis by end consumers using Garmin
products. Future studies may use more advanced HRV monitors.
Increasing the dimensions of biometric and subjective data may improve the performance of
prediction. For instance, brain waves and facial expressions can add to the inputs.
6.2.3 Diverse study objectives
This experiment compared one feature at a time. Further works can be extended to examine
the collective effects of design features in a multi-variable study. Each variable can be dissected into
a group of factors. For instance, artificial lighting can be decomposed into the lumen, product type,
color temperature, and angle. Future studies may examine the effect of each specific factor.
148

Occupants’ well-being is affected due to long periods of exposure to an unhealthy indoor
environment. Future studies may explore the relationship between short-term and long-term
influence. Many participants reported that they could not take the experiment longer than 20
minutes, which indicated a limited time to experiment. Furthermore, future studies should explore
other machine learning methods.
6.2.4 A more controlled lab setting
The air-conditioning in the lab was not working, and the indoor temperature and humidity
were not constant across the experiments. Ideally, the room temperature and moisture level should
have been maintained at the same level in order to eliminate the external factors that may affect
participants’ experience and perception. The influences of weather may be considered in future
works.
The experiment was conducted during the time of COVID-19. Participants were required to
wear masks and face covers. It is recommended that future studies test the study's replicability
without masks and face covers.
6.2.5 Comparison of real versus virtual environments
A longer-term study comparing results in real environments and results in virtual
environments could help in knowing what types of studies can be accomplished virtually versus
those that cannot.

149

6.3 Summary
People spend more than 90% of their time indoors. It is essential to investigate the impacts
of indoor design features on occupants’ experience and perception. The proposed platform
combines subjective questionnaires with objective measurements of physiological responses,
providing a more reliable measurement of occupants’ experience and perception of architectural
design features. The analysis of participants’ subjective evaluations reveals that the immersive virtual
environment can be applied to investigate occupants’ preferences of indoor design features. The
analysis of biometric data indicates that skin conductance responses can be used to measure
occupants’ emotional arousal for certain design features. The study of machine learning
demonstrates that the machine learning model can predict the users’ preferences with an accuracy of
70%. In the future, with the extension of the database, designers will be able to predict occupants’
preferred design features before the design. Moreover, designers will be able to strengthen certain
design features and thus amplify the positive influence of the design features on the human
experience.

150

Bibliography
Africa, Julia K., and Naomi A. Sachs. 2016. “Book Review: Biophilic Design: The Theory, Science
and Practice of Bringing Buildings to Life.” HERD: Health Environments Research & Design
Journal 9 (2): 176–78. https://doi.org/10.1177/1937586715618076.
Alizadehsalehi, Sepehr, Ahmad Hadavi, and Joseph Chuenhuei Huang. 2020. “From BIM to
Extended Reality in AEC Industry.” Automation in Construction 116 (August): 103254.
https://doi.org/10.1016/j.autcon.2020.103254.
Altomonte, Sergio, and Stefano Schiavon. 2013. “Occupant Satisfaction in LEED and Non-LEED
Certified Buildings.” Building and Environment 68 (October): 66–76.
https://doi.org/10.1016/J.BUILDENV.2013.06.008.
Ander, GD. 2003. Daylighting Performance and Design.
https://books.google.com/books?hl=en&lr=&id=RBMHD5Wk6LwC&oi=fnd&pg=PP15&o
ts=syWQ0p9xbT&sig=MoqHhzIqbCv9WiJ2KQcPUvDaFfo.
ARK architecture. 2020. “ARK Architects at the Technological Forefront in the Spanish
Architecture - Virtual Reality in Your New Villa | ARK.” ARK Architects. 2020. https://ark-
architects.com/ark-architects-at-the-technological-forefront-in-the-spanish-architecture-virtual-
reality/.
Baird, John C., Barbara Cassidy, and Jennifer Kurr. 1978. “Room Preference as a Function of
Architectural Features and User Activities.” Journal of Applied Psychology 63 (6): 719–27.
https://doi.org/10.1037/0021-9010.63.6.719.
Balazova, Ivana, Geo Clausen, Jens H Rindel, Torben Poulsen, and David P Wyon. 2008. “Open-
Plan Office Environments: A Laboratory Experiment to Examine the Effect of Office Noise
and Temperature on Human Perception, Comfort and Office Work Performance.” Indoor Air
2008, no. August: 17–22. https://www.nickel.odeon.dk/pdf/C101-
Indoor_Air_2008_Balazova.pdf.
Barfield, Woodrow, and Claudia Hendrix. 1995. “The Effect of Update Rate on the Sense of
Presence within Virtual Environments.” Virtual Reality 1 (1): 3–15.
https://doi.org/10.1007/BF02009709.
Benedek, Mathias, and Christian Kaernbach. 2010. “A Continuous Measure of Phasic Electrodermal
Activity.” Journal of Neuroscience Methods 190 (1): 80–91.
https://doi.org/10.1016/j.jneumeth.2010.04.028.
Bnext3D. 2017. “Play VR Apps On Phone.” 2017. https://www.youtube.com/watch?v=EHofD-
M1NUs.
Boubekri, M. 2008. Daylighting, Architecture and Health: Building Design Strategies.
https://books.google.com/books?hl=en&lr=&id=5Pc11H8VSngC&oi=fnd&pg=PP2&dq=ay
lighting,+Architecture+and+Health:+Building+Design+Strategies&ots=_jH70LKK-
S&sig=lW5hujJ96z2Txw9t4zA-_VPK5AA.
Boucsein, Wolfram. 2012. Electrodermal Activity: Second Edition. Electrodermal Activity: Second Edition.
Vol. 9781461411. Springer US. https://doi.org/10.1007/978-1-4614-1126-0.
151

Bradley, MM, and PJ Lang. 2000. “Measuring Emotion: Behavior, Feeling, and Physiology.” Cognitive
Neuroscience of Emotion.
https://books.google.com/books?hl=en&lr=&id=A2s963AzymYC&oi=fnd&pg=PA242&dq
=Bradley+%26+lang+2000&ots=mbK1jd7Dv6&sig=5oGVF1i95QBMQTFkxRidZhn1xQk.
Buchanan, Peter. 2012. “The Big Rethink - Towards a Complete Architecture.” Architectural Review,
no. January: 680–701.
https://www.proquest.com/docview/963767379/fulltext/5CFC19EB30384872PQ/1?account
id=14749.
Cacioppo, John T., Louis G. Tassinary, and Gary G. Berntson. 2016. “Strong Inference in
Psychophysiological Science.” Handbook of Psychophysiology, Fourth Edition, 3–15.
https://doi.org/10.1017/9781107415782.001.
Camm, A. John, Marek Malik, J. Thomas Bigger, Günter Breithardt, Sergio Cerutti, R. J. Cohen,
Philippe Coumel. 1996. “Heart Rate Variability. Standards of Measurement, Physiological
Interpretation, and Clinical Use. Task Force of the European Society of Cardiology and the
North American Society of Pacing and Electrophysiology.” European Heart Journal 17 (3): 354–
81. https://doi.org/10.1111/j.1542-474X.1996.tb00275.x.
Carlin, Albert S., Hunter G. Hoffman, and Suzanne Weghorst. 1997. “Virtual Reality and Tactile
Augmentation in the Treatment of Spider Phobia: A Case Report.” Behaviour Research and
Therapy 35 (2): 153–58. https://doi.org/10.1016/S0005-7967(96)00085-X.
Chamilothori, K, G Chinazzo, J. Rodrigues, E. S. Dan-Glauser, J. Wienold, and M. Andersen. 2019.
“Subjective and Physiological Responses to Façade and Sunlight Pattern Geometry in Virtual
Reality.” Building and Environment 150: 144–55. https://doi.org/10.1016/j.buildenv.2019.01.009.
Chamilothori, Kynthia, Jan Wienold, and Marilyne Andersen. 2019. “Adequacy of Immersive Virtual
Reality for the Perception of Daylit Spaces: Comparison of Real and Virtual Environments.”
LEUKOS - Journal of Illuminating Engineering Society of North America 15 (2–3): 203–26.
https://doi.org/10.1080/15502724.2017.1404918.
Chang, Chen Yen, and Ping Kun Chen. 2005. “Human Response to Window Views and Indoor
Plants in the Workplace.” HortScience 40 (5): 1354–59.
https://doi.org/10.21273/hortsci.40.5.1354.
Chen, Y., Z. Cui, and L. Hao. 2019. “Virtual Reality in Lighting Research: Comparing Physical and
Virtual Lighting Environments.” Lighting Research and Technology 51 (6): 820–37.
https://doi.org/10.1177/1477153518825387.
Clements-Croome, Derek. 2015. “Creative and Productive Workplaces: A Review.”
Https://Doi.Org/10.1080/17508975.2015.1019698 7 (4): 164–83.
https://doi.org/10.1080/17508975.2015.1019698.
Clements-Croome, Derek John, Baizhan Li, Derek Clements-Croome, and Li Baizhan. 2000.
“Productivity and Indoor Environment.” Proceedings of Healthy Buildings 1: 629.
https://www.researchgate.net/publication/279126933.
Croon, Einar M. de, Judith K Sluiter, P. Paul F.M. Kuijer, and Monique H.W. Frings-Dresen. 2005.
“The Effect of Office Concepts on Worker Health and Performance: A Systematic Review of
the Literature.” Ergonomics. https://doi.org/10.1080/00140130512331319409.
D’Oca, Simona, Tianzhen Hong, and Jared Langevin. 2018. “The Human Dimensions of Energy
152

Use in Buildings: A Review.” Renewable and Sustainable Energy Reviews 81 (January): 731–42.
https://doi.org/10.1016/J.RSER.2017.08.019.
Deniaud, Christophe, Vincent Honnet, Benoit Jeanne, and Daniel Mestre. 2015. “The Concept of
‘Presence’ as a Measure of Ecological Validity in Driving Simulators.” Journal of Interaction Science
3 (1): 1–13. https://doi.org/10.1186/s40166-015-0005-z.
Diemer, Julia, Georg W. Alpers, Henrik M. Peperkorn, Youssef Shiban, and Andreas Mühlberger.
2015. “The Impact of Perception and Presence on Emotional Reactions: A Review of Research
in Virtual Reality.” Frontiers in Psychology. Frontiers Research Foundation.
https://doi.org/10.3389/fpsyg.2015.00026.
Doulos, L., A. Tsangrassoulis, and F. Topalis. 2008. “Quantifying Energy Savings in Daylight
Responsive Systems: The Role of Dimming Electronic Ballasts.” Energy and Buildings 40 (1): 36–
50. https://doi.org/10.1016/J.ENBUILD.2007.01.019.
Edenhofer, Ottmar, Youba Sokona, Jan C Minx, Ellie Farahani, Susanne Kadner, Kristin Seyboth,
Anna Adler, et al. 2014. “Climate Change 2014 Mitigation of Climate Change Working Group
III Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate
Change Edited By.” www.cambridge.org.
Elzeyadi, I. 2011. “Daylighting-Bias and Biophilia: Quantifying the Impact of Daylighting on
Occupants Health.” School of Architecture & Allied Arts -, 1–9.
https://aceee.org/files/proceedings/2002/data/papers/SS02_Panel8_Paper08.pdf.
Elzeyadi, Ihab. 2011. “Daylighting-Bias and Biophilia: Quantifying the Impacts of Daylighting and
Views on Occupants Health Effects of Lighting on Perceptual Response in Offices View
Project High Performance K-12 Schools View Project.”
https://www.researchgate.net/publication/344361245.
Empatica. n.d. “Empatica | Medical Devices, AI and Algorithms for Remote Patient Monitoring.”
Accessed March 17, 2022. https://www.empatica.com/en-
int/index.html?utm_source=Google&utm_medium=Paid&utm_campaign=ADW_AO_BRN
D_CNV_1021&gclid=Cj0KCQjwuMuRBhCJARIsAHXdnqNXdAVDoy1r14xmiy7CKtkh8Tk
9-whe8KyWzMF7soj5cY6I5Dh3cq0aAjxtEALw_wcB.
Energy Information Administration, Us. 2021. “Electric Power Monthly with Data for June 2021.”
www.eia.gov.
Ergan, Semiha, Ahmed Radwan, Zhengbo Zou, Hua-an Tseng, and Xue Han. 2019. “Quantifying
Human Experience in Architectural Spaces with Integrated Virtual Reality and Body Sensor
Networks.” Journal of Computing in Civil Engineering 33 (2): 04018062.
https://doi.org/10.1061/(asce)cp.1943-5487.0000812.
Ergan, Semiha, Zhuoya Shi, and Xinran Yu. 2018. “Towards Quantifying Human Experience in the
Built Environment: A Crowdsourcing Based Experiment to Identify Influential Architectural
Design Features.” Journal of Building Engineering 20 (May): 51–59.
https://doi.org/10.1016/j.jobe.2018.07.004.
EURIX. 2022. “Machine Learning.” 2022. https://www.eurixgroup.com/wp-
content/uploads/2021/01/ml-e1610553826718.jpg.
Fisk, William J. 1997. “Estimates of Improved Productivity and Health from Better Indoor
Environments.” Indoor Air 7 (3): 158–72. https://doi.org/10.1111/J.1600-0668.1997.T01-1-
153

00002.X.
Fisk, WJ. 2002. “How IEQ Affects Health, Productivity.” ASHRAE Journal, 2002 - Osti.Gov.
https://www.osti.gov/servlets/purl/821649.
Foreman N. 2006. “Virtual Reality in Psychology.” Themes in Science and Technology Education 2 (1–2):
225–52. http://earthlab.uoi.gr/theste/index.php/theste/article/view/33.
Freeman, Jonathan, S. E. Avons, Don E. Pearson, and Wijnand A. Ijsselsteijn. 1999. “Effects of
Sensory Information and Prior Experience on Direct Subjective Ratings of Presence.” Presence:
Teleoperators and Virtual Environments 8 (1): 1–13. https://doi.org/10.1162/105474699566017.
Gaggioli, Andrea. 2001. “Using Virtual Reality in Experimental Psychology.” Towards Cyberpsychology:
Mind, Cognition, and Society in the Internet Age 2 (June): 157–74.
Galasiu, Anca D., and Jennifer A. Veitch. 2006. “Occupant Preferences and Satisfaction with the
Luminous Environment and Control Systems in Daylit Offices: A Literature Review.” Energy
and Buildings 38 (7): 728–42. https://doi.org/10.1016/j.enbuild.2006.03.001.
GARMIN. n.d. “Garmin Customer Support.” Accessed March 17, 2022.
https://support.garmin.com/en-US/.
Gerardi, Maryrose, Barbara Olasov Rothbaum, Kerry Ressler, Mary Heekin, and Albert Rizzo. 2008.
“Virtual Reality Exposure Therapy Using a Virtual Iraq: Case Report.” Journal of Traumatic Stress
21 (2): 209–13. https://doi.org/10.1002/jts.20331.
Géron, Aurélien. 2017. Hands-on Machine Learning with Scikit-Learn and TensorFlow : Concepts, Tools, and
Techniques to Build Intelligent Systems. O’Reilly Media.
https://books.google.com/books?hl=en&lr=&id=HHetDwAAQBAJ&oi=fnd&pg=PP1&dq
=hands+on+machine+learning+with+scikit+learn&ots=0Lqj3vloOo&sig=VpaQEE3WrfIbI
pkDY_RAHKu908E.
Glerup, H., K. Mikkelsen, L. Poulsen, E. Hass, S. Overbeck, J. Thomsen, P. Charles, and E. F.
Eriksen. 2000. “Commonly Recommended Daily Intake of Vitamin D Is Not Sufficient If
Sunlight Exposure Is Limited.” Journal of Internal Medicine 247 (2): 260–68.
https://doi.org/10.1046/J.1365-2796.2000.00595.X.
Gomez, Patrick, and Brigitta Danuser. 2004. “Affective and Physiological Responses to
Environmental Noises and Music.” International Journal of Psychophysiology 53 (2): 91–103.
https://doi.org/10.1016/j.ijpsycho.2004.02.002.
Gray, Tonia, and Carol Birrell. 2014. “Are Biophilic-Designed Site Office Buildings Linked to
Health Benefits and High Performing Occupants?” International Journal of Environmental Research
and Public Health 2014, Vol. 11, Pages 12204-12222 11 (12): 12204–22.
https://doi.org/10.3390/IJERPH111212204.
Greengard, Samuel. 2019. “Virtual Reality.” The MIT Press.
https://doi.org/10.7551/MITPRESS/11836.001.0001.
Gregg, Lynsey, and Nicholas Tarrier. 2007. “Virtual Reality in Mental Health. A Review of the
Literature.” Social Psychiatry and Psychiatric Epidemiology. https://doi.org/10.1007/s00127-007-
0173-4.
Hong, Tianzhen, Zhe Wang, Xuan Luo, and Wanni Zhang. 2020. “State-of-the-Art on Research and
Applications of Machine Learning in the Building Life Cycle.” Energy and Buildings. Elsevier.
154

https://doi.org/10.1016/j.enbuild.2020.109831.
Horr, Yousef Al, Mohammed Arif, Amit Kaushik, Ahmed Mazroei, Martha Katafygiotou, and Esam
Elsarrag. 2016. “Occupant Productivity and Office Indoor Environment Quality: A Review of
the Literature.” Building and Environment 105 (August): 369–89.
https://doi.org/10.1016/J.BUILDENV.2016.06.001.
Ihm, Pyonchan, Abderrezek Nemri, and Moncef Krarti. 2009. “Estimation of Lighting Energy
Savings from Daylighting.” Building and Environment 44 (3): 509–14.
https://doi.org/10.1016/J.BUILDENV.2008.04.016.
Insider. 2019. “VR Headset.” 2019.
https://i.insider.com/5deea18efd9db2785a050e17?width=1000&format=jpeg&auto=webp.
Izadyar, Shahram. 2017. “Autonomic Nervous System.” In Neurology Secrets: Sixth Edition, edited by
Pavol Svorc, 167–83. InTech. https://doi.org/10.1016/B978-0-323-35948-1.00013-9.
James A. Russel. 1980. “A Circumplex Model of Affect.” Journal of Personality and Social Psychology 39
(6): 1161–78. https://psycnet.apa.org/journals/psp/39/6/1161/.
Jänig, W. 1989. “Autonomic Nervous System.” Human Physiology, 333–70.
https://doi.org/10.1007/978-3-642-73831-9_16.
Jin, Yaochu, Handing Wang, and Chaoli Sun. 2021. “Introduction to Machine Learning.” In Studies
in Computational Intelligence, 975:103–45. MIT Press. https://doi.org/10.1007/978-3-030-74640-
7_4.
Jordan, Author M I, and T M Mitchell. 2015. “Machine Learning : Trends , Perspectives , and
Prospects Published by : American Association for the Advancement of Science Linked
References Are Available on JSTOR for This Article : Digitize , Preserve and Extend Access to
Science Machine Learning : Tre” 349 (6245): 255–60.
Kandel, E. R., J. H. Schwartz, T. M. Jessell, S. Siegelbaum, A. J. Hudspeth, and S. Mack. 2000.
Principles of Neural Science.
https://www.academia.edu/download/30536508/neuroscience_syllabus.pdf.
Kim, Kyung Hwan, S. W. Bang, and S. R. Kim. 2004. “Emotion Recognition System Using Short-
Term Monitoring of Physiological Signals.” Medical and Biological Engineering and Computing 42 (3):
419–27. https://doi.org/10.1007/BF02344719.
Kort, Yvonne A.W. De, Wijnand A. Ijsselsteijn, Jolien Kooijman, and Yvon Schuurmans. 2003.
“Virtual Laboratories: Comparability of Real and Virtual Environments for Environmental
Psychology.” In Presence: Teleoperators and Virtual Environments, 12:360–73.
https://doi.org/10.1162/105474603322391604.
Kosonen, R., and F. Tan. 2004. “Assessment of Productivity Loss in Air-Conditioned Buildings
Using PMV Index.” Energy and Buildings 36 (10): 987–93.
https://doi.org/10.1016/J.ENBUILD.2004.06.021.
Koutsabasis, Panayiotis, Spyros Vosinakis, Katerina Malisova, and Nikos Paparounas. 2012. “On the
Value of Virtual Worlds for Collaborative Design.” Design Studies 33 (4): 357–90.
https://doi.org/10.1016/j.destud.2011.11.004.
Krarti, Moncef. 2020. “Energy Audit of Building Systems : An Engineering Approach.” Energy Audit
of Building Systems, December. https://doi.org/10.1201/9781003011613.
155

Kreibig, Sylvia D. 2010. “Autonomic Nervous System Activity in Emotion: A Review.” Biological
Psychology. https://doi.org/10.1016/j.biopsycho.2010.03.010.
Krijn, Merel, Paul M.G. Emmelkamp, Roeline Biemond, Claudius De Wilde De Ligny, Martijn J.
Schuemie, and Charles A.P.G. Van Der Mast. 2004. “Treatment of Acrophobia in Virtual
Reality: The Role of Immersion and Presence.” Behaviour Research and Therapy 42 (2): 229–39.
https://doi.org/10.1016/S0005-7967(03)00139-6.
Kuliga, S. F., T. Thrash, R. C. Dalton, and C. Hölscher. 2015. “Virtual Reality as an Empirical
Research Tool - Exploring User Experience in a Real Building and a Corresponding Virtual
Model.” Computers, Environment and Urban Systems 54 (November): 363–75.
https://doi.org/10.1016/j.compenvurbsys.2015.09.006.
Kwallek, N, C. M. Lewis, J. W.D. Lin-Hsiao, and H. Woodson. 1996. “Effects of Nine
Monochromatic Office Interior Colors on Clerical Tasks and Worker Mood.” Color Research and
Application 21 (6): 448–58. https://doi.org/10.1002/(SICI)1520-6378(199612)21:6<448::AID-
COL7>3.0.CO;2-W.
Kwallek, Nancy. 1996. “Office Wall Color: An Assessment of Spaciousness and Preference.”
Perceptual and Motor Skills 85 (1): 49–50. https://doi.org/10.2466/pms.1996.83.1.49.
Kwallek, Nancy, Carol M. Lewis, and Ann S. Robbins. 1988. “Effects of Office Interior Color on
Workers’ Mood and Productivity.” Perceptual and Motor Skills 66 (1): 123–28.
https://doi.org/10.2466/pms.1988.66.1.123.
Lan, Li, Pawel Wargocki, and Zhiwei Lian. 2011. “Quantitative Measurement of Productivity Loss
Due to Thermal Discomfort.” In Energy and Buildings, 43:1057–62.
https://doi.org/10.1016/j.enbuild.2010.09.001.
LANG, PETER J., MARK K. GREENWALD, MARGARET M. BRADLEY, and ALFONS O.
HAMM. 1993. “Looking at Pictures: Affective, Facial, Visceral, and Behavioral Reactions.”
Psychophysiology 30 (3): 261–73. https://doi.org/10.1111/j.1469-8986.1993.tb03352.x.
Lee, Jang Hah, Brenda K. Wiederhold, Mark D. Wiederhold, Sun I. Kim, Jeonghun Ku, Wongeun
Cho, Won Yong Hahn, et al. 2003. “A Virtual Reality System for the Assessment and
Rehabilitation of the Activities of Daily Living.” In Cyberpsychology and Behavior, 6:383–88.
https://doi.org/10.1089/109493103322278763.
Levenson, Robert W. 1999. “The Intrapersonal Functions of Emotion.” Cognition and Emotion 13 (5):
481–504. https://doi.org/10.1080/026999399379159.
———. 2011. “Basic Emotion Questions.” Emotion Review 3 (4): 379–86.
https://doi.org/10.1177/1754073911410743.
———. 2014. “The Autonomic Nervous System and Emotion.” Emotion Review 6 (2): 100–112.
https://doi.org/10.1177/1754073913512003.
Levenson, Robert W., Sandy J. Lwi, Casey L. Brown, Brett Q. Ford, Marcela C. Otero, and Alice
Verstaen. 2016. “Emotion.” Handbook of Psychophysiology, Fourth Edition, January, 444–64.
https://doi.org/10.1017/9781107415782.020.
Lin, Yu Cheng, Yen Pei Chen, Huey Wen Yien, Chao Yung Huang, and Yu Chih Su. 2018.
“Integrated BIM, Game Engine and VR Technologies for Healthcare Design: A Case Study in
Cancer Hospital.” Advanced Engineering Informatics 36 (April): 130–45.
https://doi.org/10.1016/J.AEI.2018.03.005.
156

Logadóttir, Ásta, J. Christoffersen, and S. A. Fotios. 2011. “Investigating the Use of an Adjustment
Task to Set the Preferred Illuminance in a Workplace Environment.” Lighting Research and
Technology 43 (4): 403–22. https://doi.org/10.1177/1477153511400971.
Lohr, Virginia I., Caroline H. Pearson-Mims, and Georgia K. Goodwin. 1996. “Interior Plants May
Improve Worker Productivity and Reduce Stress in a Windowless Environment.” Journal of
Environmental Horticulture 14 (2): 97–100. https://doi.org/10.24266/0738-2898-14.2.97.
Loomis, Jack M., James J. Blascovich, and Andrew C. Beall. 1999. “Immersive Virtual Environment
Technology as a Basic Research Tool in Psychology.” Behavior Research Methods, Instruments, and
Computers 31 (4): 557–64. https://doi.org/10.3758/BF03200735.
MacKerron, George, and Susana Mourato. 2013. “Happiness Is Greater in Natural Environments.”
Global Environmental Change 23 (5): 992–1000. https://doi.org/10.1016/j.gloenvcha.2013.03.010.
Mahnke, Frank H. 1996. Color, Environment, and Human Response: An Interdisciplinary Understanding.
https://books.google.com/books?hl=en&lr=&id=fAsm_3cXISAC&oi=fnd&pg=PP13&ots=
2u434cS2C0&sig=Q2Ny0PqOKJNqwoumrsrZh1U0CAg.
Majboroda, Sergej. n.d. “Landscape HDRI.” Poly Haven. Accessed March 17, 2022.
https://cdn.polyhaven.com/asset_img/primary/hilly_terrain_01.png?height=780.
Milgram, Paul, Haruo Takemura, Akira Utsumi, and Fumio Kishino. 1995. “Augmented Reality: A
Class of Displays on the Reality-Virtuality Continuum.” Telemanipulator and Telepresence
Technologies 2351 (21): 282–92. https://doi.org/10.1117/12.197321.
Müller, A., & Guido, S. 2018. Introduction to Machine Learning with Python: A Guide for Data Scientists 1st
Edition. O’Reilly Media. Journal of Chemical Information and Modeling. Vol. 53.
https://books.google.com/books?hl=en&lr=&id=1-
4lDQAAQBAJ&oi=fnd&pg=PP1&dq=introduction+to+machine+learning+with+python&o
ts=28kRIHPMY_&sig=qX5ay5NYPolKM_4DkzLjO83WYaQ.
Naz, K, and H Epps. 2004. “Relationship between Color and Emotion: A Study of College
Students.” College Student J 38 (3): 396–405.
https://www.academia.edu/download/48662263/Kaya_Epps_2004b.pdf.
Oculus. 2020. “Oculus Quest 2: Our Most Advanced New All-in-One VR Headset | Oculus.” 2020.
https://www.oculus.com/quest-
2/?utm_source=gg&utm_medium=ps&utm_campaign=11065328390&utm_term=oculus
quest
2&utm_content=560568920808&utm_ad=106798706845&utm_location=9073456&utm_locat
ion2&utm_placement=kwd-563298318914&utm_device=c&utm_matchtype=e&utm_fe.
Paljic, Alexis. 2017. “Ecological Validity of Virtual Reality: Three Use Cases.” In Lecture Notes in
Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in
Bioinformatics), 10590 LNCS:301–10. Springer, Cham. https://doi.org/10.1007/978-3-319-
70742-6_28.
Palomba, Daniela, Alessandro Angrilli, and Alessio Mini. 1997. “Visual Evoked Potentials, Heart
Rate Responses and Memory to Emotional Pictorial Stimuli.” International Journal of
Psychophysiology 27 (1): 55–67. https://doi.org/10.1016/S0167-8760(97)00751-4.
Parsons, Thomas D. 2011. “Neuropsychological Assessment Using Virtual Environments:
Enhanced Assessment Technology for Improved Ecological Validity.” Studies in Computational
157

Intelligence 337: 271–89. https://doi.org/10.1007/978-3-642-17824-5_13.
———. 2015. “Virtual Reality for Enhanced Ecological Validity and Experimental Control in the
Clinical, Affective and Social Neurosciences.” Frontiers in Human Neuroscience 9 (DEC).
https://doi.org/10.3389/fnhum.2015.00660.
Paul, Warren L., and Peter A. Taylor. 2008. “A Comparison of Occupant Comfort and Satisfaction
between a Green Building and a Conventional Building.” Building and Environment 43 (11):
1858–70. https://doi.org/10.1016/J.BUILDENV.2007.11.006.
Paulhus, DL, and S Vazire. 2007. The Self-Report Method. Books.Google.Com.
https://books.google.com/books?hl=en&lr=&id=XHwS3PU6uroC&oi=fnd&pg=PA224&dq
=self+report&ots=JDLhUYQO_B&sig=vsaVGXMZiLoNSi0NXZx44yxHSpk.
Radwan, A., and S. Ergan. 2017. “Quantifying Human Experience in Interior Architectural Spaces.”
In Congress on Computing in Civil Engineering, Proceedings, 2017-June:373–80. American Society of
Civil Engineers (ASCE). https://doi.org/10.1061/9780784480830.046.
Rashid, Mahbub, and Craig Zimring. 2008. “A Review of the Empirical Literature on the
Relationships Between Indoor Environment and Stress in Health Care and Office Settings
Problems and Prospects of Sharing Evidence.” Environment and Behavior 40.
https://doi.org/10.1177/0013916507311550.
Redlich, Carrie A., Judy Sparer, and Mark R. Cullen. 1997. “Sick-Building Syndrome.” The Lancet 349
(9057): 1013–16. https://doi.org/10.1016/S0140-6736(96)07220-0.
Robinson, S, and J Pallasmaa. 2015. Mind in Architecture: Neuroscience, Embodiment, and the Future of
Design.
https://books.google.com/books?hl=en&lr=&id=9ohLCAAAQBAJ&oi=fnd&pg=PR5&dq=
Mind+in+Architecture+:+Neuroscience,+Embodiment,+and+the+Future+of+Design&ots=
Vld1heaEg8&sig=Bp_VOqv8Wj0hAgL0FFNGCygDiso.
Sadalla, Edward K., and Virgil L. Sheets. 1993. “Symbolism in Building Materials: Self-Presentational
and Cognitive Components.” Environment and Behavior 25 (2): 155–80.
https://doi.org/10.1177/0013916593252001.
Sanchez-Vives, Maria V., and Mel Slater. 2005. “From Presence to Consciousness through Virtual
Reality.” Nature Reviews Neuroscience. https://doi.org/10.1038/nrn1651.
Sarah Robinson, and Juhani Pallasmaa. 2015. Mind in Architecture : Neuroscience, Embodiment, and the
Future of Design. MIT Press.
https://ebookcentral.proquest.com/lib/socal/detail.action?docID=3339978.
Sarbu, Ioan, and Calin Sebarchievici. 2013. “Aspects of Indoor Environmental Quality Assessment
in Buildings.” Energy and Buildings 60 (May): 410–19.
https://doi.org/10.1016/J.ENBUILD.2013.02.005.
Schmuckler, Mark A. 2001. “What Is Ecological Validity? A Dimensional Analysis.” Infancy 2 (4):
419–36. https://doi.org/10.1207/S15327078IN0204_02.
Shusterman, Vladimir, and Ofer Barnea. 1995. “Skin-Temperature Variability Indicates Vascular
Sympathetic Activity.” In Proceedings - IEEE Convention of Electrical & Electronics Engineers in Israel.
https://doi.org/10.1109/eeis.1995.514158.
Singaravel, Sundaravelpandian, Johan Suykens, and Philipp Geyer. 2018. “Deep-Learning Neural-
158

Network Architectures and Methods: Using Component-Based Models in Building-Design
Energy Prediction.” Advanced Engineering Informatics 38 (October): 81–90.
https://doi.org/10.1016/j.aei.2018.06.004.
Sink, DS. 1985. “Productivity Management: Planning, Maesurement and Evaluation, Control and
Improvement.”
http://repo.unikadelasalle.ac.id/index.php?author=%22SINK%2C+D.+Scott%22&search=Se
arch.
Slater, Mel, Beau Lotto, Maria Marta Arnold, and Maria V. Sanchez-Vives. 2009. “How We
Experience Immersive Virtual Environments: The Concept of Presence and Its Measurement.”
Anuario de Psicologia. https://doi.org/10.1344/anuario.any.volum.numero.
Slater, Mel, and Sylvia Wilbur. 1997. “A Framework for Immersive Virtual Environments (FIVE):
Speculations on the Role of Presence in Virtual Environments.” Presence: Teleoperators and Virtual
Environments 6 (6): 603–16. https://doi.org/10.1162/pres.1997.6.6.603.
Tseng, Kevin C., Do Thi Ngoc Giau, and Po Hsin Huang. 2017. “A Systematic Review of the
Potential Application of Virtual Reality within a User Pre-Occupancy Evaluation.” In Lecture
Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in
Bioinformatics), 10279 LNCS:612–20. Springer Verlag. https://doi.org/10.1007/978-3-319-
58700-4_50.
U.S. EPA. n.d. “Indoor Air Quality | US EPA.” Accessed January 19, 2022.
https://www.epa.gov/report-environment/indoor-air-quality#note1.
Unity Technologies. n.d. “High Definition Render Pipeline Overview | High Definition RP |
7.1.8.” Docs.Unity3d.Com. Accessed November 19, 2021.
https://docs.unity3d.com/Packages/com.unity.render-pipelines.high-
definition@12.1/manual/index.html.
VanderPlas, Jake. 2016. Python Data Science Handbook: Essential Tools for Working with Data. O’Reilly.
https://books.google.com/books?hl=en&lr=&id=xYmNDQAAQBAJ&oi=fnd&pg=PR2&d
q=+python+data+science+handbook&ots=XrbNp6jg5N&sig=kwQOz8gACUltKFyUbYMS
Hwh2Jdg.
Vartanian, Oshin, Gorka Navarrete, Anjan Chatterjee, Lars Brorson Fich, Jose Luis Gonzalez-Mora,
Helmut Leder, Cristián Modroño, Marcos Nadal, Nicolai Rostrup, and Martin Skov. 2015.
“Architectural Design and the Brain: Effects of Ceiling Height and Perceived Enclosure on
Beauty Judgments and Approach-Avoidance Decisions.” Journal of Environmental Psychology 41:
10–18. https://doi.org/10.1016/j.jenvp.2014.11.006.
Visbox. n.d. “Multi-Screen Applications.” Accessed March 16, 2022.
http://www.visbox.com/wp/wp-content/uploads/2019/03/NASA-crop-2000.jpg.
Vrana, Scott R., and David Rollock. 2002. “The Role of Ethnicity, Gender, Emotional Content, and
Contextual Differences in Physiological, Expressive, and Self-Reported Emotional Responses
to Imagery.” Cognition and Emotion 16 (1): 165–92.
https://doi.org/10.1080/02699930143000185.
Wang, Xiao Wei, Dan Nie, and Bao Liang Lu. 2014. “Emotional State Classification from EEG
Data Using Machine Learning Approach.” Neurocomputing 129: 94–106.
https://doi.org/10.1016/j.neucom.2013.06.046.
159

Westerdahl, Börje, Kaj Suneson, Claes Wernemyr, Mattias Roupé, Mikael Johansson, and Carl
Martin Allwood. 2006. “Users’ Evaluation of a Virtual Reality Architectural Model Compared
with the Experience of the Completed Building.” Automation in Construction 15 (2): 150–65.
https://doi.org/10.1016/j.autcon.2005.02.010.
Wilson, Christopher J., and Alessandro Soranzo. 2015. “The Use of Virtual Reality in Psychology: A
Case Study in Visual Perception.” Computational and Mathematical Methods in Medicine.
https://doi.org/10.1155/2015/151702.
Wilson, Edward O. 2018. “Biophilia and the Conservation Ethic.” In Evolutionary Perspectives on
Environmental Problems, 250–58. Routledge. https://doi.org/10.4324/9780203792650-18.
Yeom, Dongwoo, Joon Ho Choi, and Sin Hwa Kang. 2019. “Investigation of the Physiological
Differences in the Immersive Virtual Reality Environment and Real Indoor Environment:
Focused on Skin Temperature and Thermal Sensation.” Building and Environment 154 (May): 44–
54. https://doi.org/10.1016/j.buildenv.2019.03.013.
Zhang, Yuxuan, Hexu Liu, Shih Chung Kang, and Mohamed Al-Hussein. 2020. “Virtual Reality
Applications for the Built Environment: Research Trends and Opportunities.” Automation in
Construction. Elsevier. https://doi.org/10.1016/j.autcon.2020.103311.
Zou, Hao, Nan Li, and Lijun Cao. 2017. “Emotional Response–Based Approach for Assessing the
Sense of Presence of Subjects in Virtual Building Evacuation Studies.” Journal of Computing in
Civil Engineering 31 (5): 04017028. https://doi.org/10.1061/(asce)cp.1943-5487.0000679.

160

Appendices
Appendix A 14 design variables
Daylight

Artificial lighting

161

Low illuminance level

High illuminance level

162

Green view

No green view

163

No interior plants

Plants

164

Bright color

Red color

165

Fabric texture

Wood texture

166

Low ceiling height (2.8m)

High ceiling height (3.8m)

167

Appendix B Participants’ responses to the in-VR questionnaire
num ID
Event.Name satisfaction motivation stress concentrate
1 SA1
baseline
1 SA1
arti lighting -1 -1 0 1
1 SA1
daylight 1 -1 0 2
1 SA1
highIllu 0 -2 -1 -1
1 SA1
lowIllu 1 0 1 1
1 SA1
cityView -1 -1 -1 -1
1 SA1
greenView 1 -1 2 2
1 SA1
Plants 1 0 1 1
1 SA1
noPlants -1 -2 0 -1
1 SA1
bright -1 -1 0 -1
1 SA1
red -2 -1 -2 -1
1 SA1
fabric -2 -1 -1 -1
1 SA1
wood -1 -1 -1 -2
1 SA1
highCeiling 1 -1 0 1
1 SA1
lowCeiling 0 0 -1 1
2 SB1
baseline
2 SB1
arti lighting 1 0 -1 1
2 SB1
daylight 2 0 1 1
2 SB1
highIllu 2 1 2 -1
2 SB1
lowIllu 1 0 0 1
2 SB1
cityView 2 1 2 1
2 SB1
greenView 2 1 2 0
2 SB1
Plants 2 1 2 1
2 SB1
noPlants 1 0 2 0
2 SB1
bright 0 -1 1 -2
2 SB1
red 1 1 1 1
2 SB1
fabric 1 0 0 0
2 SB1
wood 1 1 2 -1
2 SB1
highCeiling 1 1 1 1
2 SB1
lowCeiling -2 -2 -2 -2
3 SB2
baseline
3 SB2
arti lighting 1 2 1 1
3 SB2
daylight 0 1 1 0
3 SB2
highIllu 1 1 0 1
3 SB2
lowIllu 0 1 1 0
168

3 SB2
cityView 1 1 1 0
3 SB2
greenView 1 2 2 1
3 SB2
Plants 1 2 1 1
3 SB2
noPlants 1 0 1 0
3 SB2
bright -1 -1 -1 -1
3 SB2
red 1 1 0 0
3 SB2
fabric 2 2 2 1
3 SB2
wood 2 2 1 1
3 SB2
highCeiling 2 2 1 1
3 SB2
lowCeiling -1 -1 -1 -1
4 SB3
baseline
4 SB3
arti lighting -1 -1 -1 2
4 SB3
daylight 1 1 1 2
4 SB3
highIllu 1 1 2 2
4 SB3
lowIllu 0 1 0 1
4 SB3
cityView 1 2 1 2
4 SB3
greenView 2 2 1 2
4 SB3
Plants 1 1 1 1
4 SB3
noPlants 1 0 0 0
4 SB3
bright 1 1 -1 -2
4 SB3
red 0 0 -1 0
4 SB3
fabric 1 1 2 1
4 SB3
wood -1 -1 -1 -1
4 SB3
highCeiling 1 1 1 1
4 SB3
lowCeiling -2 -2 -2 -1
5 SC1
baseline
5 SC1
arti lighting 1 0 -1 0
5 SC1
daylight 1 1 1
5 SC1
highIllu 1 1 0 1
5 SC1
lowIllu -1 -1 -2 -1
5 SC1
cityView 0 0 0 0
5 SC1
greenView 2 2 1 1
5 SC1
Plants 0 0 1 0
5 SC1
noPlants 0 -1 0 0
5 SC1
red -1 -1 -2 -2
5 SC1
bright 1 1 1 1
5 SC1
wood -1 -1 -1 -1
5 SC1
fabric -1 -1 -1 -1
5 SC1
highCeiling 1 0 1 0
5 SC1
lowCeiling 0 0 0 0
6 SD1
baseline
169

6 SD1
daylight 1 2 1 1
6 SD1
arti lighting 1 1 0 1
6 SD1
lowIllu 1 2 1 2
6 SD1
highIllu -1 -1 -1 -1
6 SD1
greenView 2 2 2 2
6 SD1
cityView 1 -1 0 1
6 SD1
noPlants 1 1 0 1
6 SD1
Plants 1 1 0 -1
6 SD1
bright 2 2 2 2
6 SD1
red -1 -1 -1 -1
6 SD1
fabric 1 0 0 0
6 SD1
wood 2 2 1 2
6 SD1
lowCeiling -1 -1 1 -1
6 SD1
highCeiling 1 1 1 1
7 SE1
baseline
7 SE1
arti lighting 2 2 1 1
7 SE1
daylight 2 2 2 1
7 SE1
highIllu 2 2 2 2
7 SE1
lowIllu 2 2 2 2
7 SE1
cityView 1 1 1 0
7 SE1
greenView 2 2 2 2
7 SE1
Plants 1 1 2 2
7 SE1
noPlants 1 1 0 1
7 SE1
red -1 -1 -1 -2
7 SE1
bright 0 1 0 1
7 SE1
wood -1 -1 -1 -1
7 SE1
fabric -1 -1 0 0
7 SE1
highCeiling 1 1 2 2
7 SE1
lowCeiling 1 2 1 2
8 SE2
baseline
8 SE2
arti lighting 2 2 1 1
8 SE2
daylight 2 2 1 2
8 SE2
highIllu 2 1 1 2
8 SE2
lowIllu 2 2 1 2
8 SE2
cityView 2 2 2 2
8 SE2
greenView 2 2 2 2
8 SE2
Plants 2 2 2 2
8 SE2
noPlants 1 1 1 1
8 SE2
red -1 -2 -2 -1
8 SE2
bright 2 1 1 1
8 SE2
wood -1 -2 -2 -1
170

8 SE2
fabric 2 1 1 1
8 SE2
highCeiling 2 1 2 1
8 SE2
lowCeiling 0 -1 -1 -1
9 SE3
baseline
9 SE3
arti lighting 1 1 1 1
9 SE3
daylight 2 2 2 2
9 SE3
highIllu 2 2 2 2
9 SE3
lowIllu -1 -1 -1 0
9 SE3
cityView 0 0 0 0
9 SE3
greenView 2 2 2 2
9 SE3
Plants 1 1 1 1
9 SE3
noPlants 0 0 1 1
9 SE3
red -1 -1 -1 -1
9 SE3
bright 1 1 1 1
9 SE3
wood -1 0 -1 0
9 SE3
fabric 1 1 1 1
9 SE3
highCeiling 0 0 1 0
9 SE3
LowCeilingMissing
10 SE4
baseline
10 SE4
arti lighting 1 0 -1 1
10 SE4
daylight 2 2 2 2
10 SE4
highIllu 1 1 2 1
10 SE4
lowIllu 1 0 0 0
10 SE4
cityView 2 1 2 1
10 SE4
greenView 2 2 2 1
10 SE4
Plants 2 2 1 2
10 SE4
noPlants 1 1 1 1
10 SE4
red 0 0 -2 0
10 SE4
bright 2 2 2 2
10 SE4
wood 0 0 -1 0
10 SE4
fabric 1 0 -1 1
10 SE4
highCeiling 2 2 2 2
10 SE4
lowCeiling -1 -1 -2 -1
11 SE5
baseline
11 SE5
arti lighting 0 -1 0 -1
11 SE5
daylight 1 2 -1 1
11 SE5
highIllu 2 2 -1 2
11 SE5
lowIllu -1 -2 0 -1
11 SE5
cityView 0 -1 1 1
11 SE5
greenView 2 2 0 2
11 SE5
Plants 2 2 1 2
171

11 SE5
noPlants -1 0 -1 1
11 SE5
red -1 -2 -1 -2
11 SE5
bright 1 1 -1 1
11 SE5
wood 1 1 0 1
11 SE5
fabric 0 -1 1 -1
11 SE5
highCeiling 1 2 1 2
11 SE5
lowCeiling 0 0 1 0
12 SE6
baseline
12 SE6
daylight 0 -1 -1 2
12 SE6
arti lighting 1 0 1 1
12 SE6
lowIllu -1 -2 -1 0
12 SE6
highIllu 2 2 0 2
12 SE6
greenView 2 2 2 2
12 SE6
cityView 1 1 1 2
12 SE6
noPlants 0 0 1 -1
12 SE6
Plants 1 2 1 1
12 SE6
bright 0 -1 -1 1
12 SE6
red 0 -2 2 -2
12 SE6
fabric 1 1 2 2
12 SE6
wood 1 -1 2 0
12 SE6
lowCeiling -1 -2 -2 -2
12 SE6
highCeiling -1 -1 -1 -1
13 SF1
baseline
13 SF1
arti lighting -1 -1 0 -2
13 SF1
daylight 0 0 -1 -1
13 SF1
lowIllu 0 0 -1 -1
13 SF1
highIllu 1 0 0 0
13 SF1
greenView 1 1 1 0
13 SF1
cityView 1 0 1 0
13 SF1
noPlants 0 1 0 -1
13 SF1
Plants 1 1 0 -1
13 SF1
bright -1 -1 -1 -1
13 SF1
red -1 -1 -1 -1
13 SF1
fabric 0 0 0 0
13 SF1
wood 0 -1 -1 -1
13 SF1
lowCeiling 0 -1 0 -1
13 SF1
highCeiling 0 -1 0 -1
14 SF2
baseline
14 SF2
daylight 2 2 2 1
14 SF2
arti lighting 1 2 2 2
14 SF2
lowIllu 1 1 2 1
172

14 SF2
highIllu -1 -1 -1 -1
14 SF2
cityView 1 1 0 0
14 SF2
greenView 2 2 2 1
14 SF2
noPlants 1 1 2 1
14 SF2
Plants 2 2 2 2
14 SF2
bright -1 -1 -1 -1
14 SF2
red -1 -1 -1 -2
14 SF2
fabric 1 1 2 1
14 SF2
wood 1 1 2 1
14 SF2
lowCeiling 2 2 2 2
14 SF2
highCeiling 2 2 2 2
15 SF3
baseline
15 SF3
arti lighting 1 1 0 1
15 SF3
daylight 0 1 0 1
15 SF3
highIllu 0 0 -1 1
15 SF3
lowIllu 1 1 0 1
15 SF3
cityView 1 1 0 1
15 SF3
greenView 1 2 1 -1
15 SF3
Plants 1 1 1 1
15 SF3
noPlants 0 1 0 1
15 SF3
red -1 0 -1 0
15 SF3
bright 1 1 1 1
15 SF3
wood -1 0 -1 0
15 SF3
fabric 0 -1 -1 -1
15 SF3
highCeiling 0 1 1 1
15 SF3
lowCeiling -1 -1 0 1
16 SF4
baseline
16 SF4
daylight 1 1 1 2
16 SF4
arti lighting -1 -2 -1 -2
16 SF4
lowIllu 0 1 1 1
16 SF4
highIllu 2 1 1 2
16 SF4
greenView 2 2 2 2
16 SF4
cityView 0 -1 -1 1
16 SF4
noPlants -1 -1 -1 -1
16 SF4
Plants 1 1 0 1
16 SF4
bright 2 1 1 2
16 SF4
red -1 -2 -1 -2
16 SF4
fabric 1 1 0 1
16 SF4
wood 2 1 1 1
16 SF4
lowCeiling -1 -1 -1 -2
16 SF4
highCeiling 2 2 0 1
173

17 SF5
baseline
17 SF5
arti lighting 1 0 0 1
17 SF5
daylight 2 2 2 1
17 SF5
lowIllu -1 -2 0 -1
17 SF5
highIllu 1 2 -1 1
17 SF5
greenView 0 0 0 -1
17 SF5
cityView 0 1 -1 0
17 SF5
noPlants 1 2 1 2
17 SF5
Plants 2 2 2 2
17 SF5
bright 1 1 -1 0
17 SF5
red -2 -2 -2 -2
17 SF5
fabric 0 0 0 0
17 SF5
wood 2 2 -1 1
17 SF5
lowCeiling 0 0 0 0
17 SF5
highCeiling -1 -1 0 0
18 SG1
baseline
18 SG1
daylight 1 2 2 1
18 SG1
arti lighting 0 -2 1 -1
18 SG1
lowIllu 1 0 0 1
18 SG1
highIllu -1 -1 -1 -1
18 SG1
greenView 2 1 2 1
18 SG1
cityView 1 0 1 1
18 SG1
noPlants 0 -1 -1 -1
18 SG1
Plants 1 1 0 0
18 SG1
bright 1 1 0 1
18 SG1
red -2 -2 -2 -2
18 SG1
fabric 1 0 0 -1
18 SG1
wood 1 0 0 -1
18 SG1
lowCeiling 1 0 0 -1
18 SG1
highCeiling 1 0 1 1
19 SH1
baseline
19 SH1
arti lighting 2 1 -1 0
19 SH1
daylight 1 1 1 1
19 SH1
lowIllu 1 0 0 -1
19 SH1
highIllu 1 1 0 1
19 SH1
cityView 1 -1 -1 1
19 SH1
greenView 2 1 0 1
19 SH1
noPlants 0 -1 -1 -1
19 SH1
Plants 1 1 1 1
19 SH1
red 0 -1 -2 -1
19 SH1
bright 1 0 0 0
174

19 SH1
wood 0 -1 1 1
19 SH1
fabric -1 -1 -1 -1
19 SH1
highCeiling 1 -1 1 1
19 SH1
lowCeiling 1 1 0 1
20 SH2
baseline
20 SH2
arti lighting -1 -1 0 0
20 SH2
daylight 1 1 2 1
20 SH2
highIllu 1 1 2 1
20 SH2
lowIllu 0 -1 0 0
20 SH2
cityView 0 0 0 0
20 SH2
greenView 2 1 2 1
20 SH2
Plants 2 2 1 1
20 SH2
noPlants -1 -1 -1 -1
20 SH2
red -2 -2 -2 -2
20 SH2
bright -1 -1 -1 -1
20 SH2
wood -1 -1 0 0
20 SH2
fabric -1 -1 0 0
20 SH2
highCeiling 1 1 0 1
20 SH2
lowCeiling -1 -1 -1 0
21 SH3
baseline
21 SH3
arti lighting 0 1 0 1
21 SH3
daylight 2 1 1 -1
21 SH3
highIllu 1 0 -1 0
21 SH3
lowIllu -1 -1 0 -2
21 SH3
cityView 0 1 0 1
21 SH3
greenView 1 -2 2 -1
21 SH3
Plants 1 1 2 1
21 SH3
noPlants 0 -1 0 1
21 SH3
red -1 -1 2 -1
21 SH3
bright 1 0 1 1
21 SH3
wood 0 -1 1 -2
21 SH3
fabric -1 -1 0 0
21 SH3
highCeiling 0 1 0 -1
21 SH3
lowCeiling -2 -2 1 -1
22 SH4
baseline
22 SH4
daylight 2 2 2 1
22 SH4
arti lighting 1 -2 -1 2
22 SH4
lowIllu 1 1 0 1
22 SH4
highIllu 0 -1 -1 2
22 SH4
greenView 2 2 2 1
22 SH4
cityView 1 1 0 1
175

22 SH4
noPlants 1 1 0 1
22 SH4
Plants 2 2 2 2
22 SH4
bright 1 2 -2 2
22 SH4
red 0 -2 2 0
22 SH4
fabric 2 0 2 1
22 SH4
wood 2 2 2 2
22 SH4
lowCeiling -2 -2 -2 -1
22 SH4
highCeiling 1 1 0 1
23 SH5
baseline
23 SH5
daylight 0 1 1 1
23 SH5
arti lighting -1 -1 -1 -1
23 SH5
lowIllu -2 -1 -1 -1
23 SH5
highIllu 1 1 0 1
23 SH5
greenView 2 1 -2 2
23 SH5
cityView 0 0 0 0
23 SH5
noPlants 0 0 0 0
23 SH5
Plants 1 1 1 1
23 SH5
bright 1 1 1 1
23 SH5
red -2 -2 -2 -2
23 SH5
fabric -1 -1 -1 -2
23 SH5
wood 1 0 0 -1
23 SH5
lowCeiling 0 0 -1 -1
23 SH5
highCeiling 1 1 0 0
24 SI1
baseline
24 SI1
daylight 1 1 0 1
24 SI1
arti lighting 0 0 -1 0
24 SI1
lowIllu 2 1 1 2
24 SI1
highIllu 1 1 1 1
24 SI1
greenView 2 2 1 2
24 SI1
cityView 0 0 -1 0
24 SI1
noPlants 0 0 -1 0
24 SI1
Plants 1 1 1 1
24 SI1
red 0 -1 1 -1
24 SI1
bright 0 1 0 0
24 SI1
fabric 0 -1 0 -1
24 SI1
wood 0 0 -1 0
24 SI1
lowCeiling 0 0 0 0
24 SI1
highCeiling 2 1 1 0
25 SI2
baseline
25 SI2
daylight 1 1 1 -1
25 SI2
arti lighting 2 -2 0 -2
176

25 SI2
highIllu 2 2 2 2
25 SI2
lowIllu -1 -1 0 0
25 SI2
cityView 2 2 1 2
25 SI2
greenView 2 2 2 1
25 SI2
noPlants 1 1 -1 1
25 SI2
Plants 2 2 2 1
25 SI2
bright 1 2 -1 2
25 SI2
red -2 -2 -2 -2
25 SI2
wood -1 -1 -1 0
25 SI2
fabric -1 -1 1 0
25 SI2
highCeiling 1 2 2 2
25 SI2
lowCeiling -2 0 2 -1
26 SI3
baseline
26 SI3
arti lighting 2 2 1 2
26 SI3
daylight 2 2 2 1
26 SI3
lowIllu 1 2 2 2
26 SI3
highIllu 0 1 0 1
26 SI3
cityView 1 1 1 1
26 SI3
greenView 2 1 2 1
26 SI3
Plants 0 1 1 0
26 SI3
noPlants 1 2 0 1
26 SI3
bright 0 0 1 0
26 SI3
red -1 -1 0 -1
26 SI3
wood -1 0 0 0
26 SI3
fabric 2 1 -1 2
26 SI3
lowCeiling 2 2 -1 1
26 SI3
highCeiling 2 2 -1 1
27 SJ1
baseline
27 SJ1
arti lighting 2 1 1 0
27 SJ1
daylight 2 2 2 2
27 SJ1
highIllu 1 0 1 0
27 SJ1
lowIllu 2 1 2 2
27 SJ1
cityView 1 1 1 1
27 SJ1
greenView 2 2 2 2
27 SJ1
Plants 0 0 0 1
27 SJ1
noPlants 1 0 0 0
27 SJ1
red -1 -2 -1 -2
27 SJ1
bright 0 0 0 0
27 SJ1
wood 2 1 1 -1
27 SJ1
fabric 1 1 1 1
27 SJ1
highCeiling 1 1 1 1
177

27 SJ1
lowCeiling -2 -2 -2 0
28 SK1
baseline
28 SK1
daylight 1 0 0 1
28 SK1
arti lighting 0 -1 1 0
28 SK1
highIllu 1 1 -1 1
28 SK1
lowIllu 0 0 1 0
28 SK1
cityView 1 0 0 -1
28 SK1
greenView 2 1 0 1
28 SK1
Plants 1 1 0 0
28 SK1
noPlants 0 1 -1 1
28 SK1
bright 1 1 -1 1
28 SK1
red -1 -1 0 -2
28 SK1
fabric 0 0 0 0
28 SK1
wood 0 0 0 0
28 SK1
lowCeiling -1 0 0 -1
28 SK1
highCeiling 1 1 -1 1
29 SK2
baseline
29 SK2
daylight 1 1 0 1
29 SK2
arti lighting -1 -1 -1 -1
29 SK2
lowIllu 0 0 -1 0
29 SK2
highIllu 1 1 1 1
29 SK2
greenView 2 2 2 1
29 SK2
cityView 0 -1 0 0
29 SK2
Plants 1 1 1 1
29 SK2
noPlants 0 0 -1 1
29 SK2
bright -1 -1 -1 -2
29 SK2
red -2 -2 -2 -2
29 SK2
wood 1 1 1 1
29 SK2
fabric -1 0 -1 -1
29 SK2
lowCeiling -1 -1 -1 -1
29 SK2
highCeiling 1 1 1 1
30 SK3
baseline
30 SK3
arti lighting -1 0 0 1
30 SK3
daylight 1 1 1 2
30 SK3
highIllu 1 2 2 1
30 SK3
lowIllu 0 0 0 1
30 SK3
cityView 1 1 0 1
30 SK3
greenView 2 2 2 2
30 SK3
Plants 1 1 1 1
30 SK3
noPlants 1 0 1 0
30 SK3
red -2 -2 -2 -2
178

30 SK3
bright 2 2 2 2
30 SK3
wood 0 0 -1 0
30 SK3
fabric -1 -1 0 0
30 SK3
highCeiling 1 0 1 1
30 SK3
lowCeiling -2 -2 -2 -1
Appendix C Questionnaire of the presence
num ID To what
extent did
you have a
sense of
being in the
office? (not
at all -2, -1,
0, 1, 2 very
much so)
How much
time did you
feel the office
is 'real', and
you forgot the
'real world'?
(Never -2, -1,
0, 1, 2 almost
all the time)
When you
think back of
your
experience,
do you think
the office as
images that
you saw, or
more as
somewhere
that you visit?
(only as
images -2, -1,
0, 1, 2
somewhere I
visited)
Adding what factor can
improve your sense of
immersion? (e.g., sound,
avatar (have a body), more
interaction with the scene,
better rendering, add npc,
wireless etc.?)
Did you
have
motion
sickness?
1 SA1 0 -1 2 sound, better rendering no
2 SB1 0 -1 1 npc, sound no
3 SB2 1 -1 0 sound yes
4 SB3 1 0 2 more interaction with the
scene
no
5 SC1 1 -1 0 sound yes
6 SD1 1 1 2 sound, better rendering no
7 SE1 2 1 2 Better rendering no
8 SE2 1 1 2 more realistic texture no
9 SE3 1 1 1 npc no
10 SE4 2 1 1 Better rendering, sound no
11 SE5 2 2 0 sound, more interavtion yes
12 SE6 2 0 0 sound, avatar, more
interaction
no
13 SF1 -1 0 0 sound, smell, texture yes
14 SF2 2 1 1 Sound, Human figures,
Interaction with the human
figures
no
15 SF3 2 1 1 maybe some
shadows/dynamic shadows
no
16 SF4 2 2 2 npc yes
179

17 SF5 2 0 1 better rendering yes
18 SG1 1 1 2 Skip the movements Yes
19 SH1 2 1 2 none Yes
20 SH2 2 1 2 Having other people there,
sounds
No
21 SH3 0 -1 1 Want to sit. Computers on
for rendering
No
22 SH4 2 0 1 sound, more interaction with
scene, wireless, live
(movable) object
no
23 SH5 1 1 2 sound yes
24 SI1 2 1 2 sound in the scene, and may
be plants moving if there is
natural ventilation
no
25 SI2 2 2 2 easier doors yes
26 SI3 1 2 2 occupants, cloth,office stuff,
mirror,animation
yes
27 SJ1 1 1 1 Sound, some human models no
28 SK1 1 1 2 sound, add npc no
29 SK2 2 2 1 people no
30 SK3 1 -1 1 rendering, movement, people no
180

Appendix D Experiment Procedure

181

Appendix E Integrated data
Column 1 - 8
num ID Event.Nr Event.Name CDA.nSCR CDA.Latency CDA.AmpSum CDA.SCR
Column 9 - 16
CDA.IS
CR
CDA.Phasic
Max
CDA.To
nic
TTP.nS
CR
TTP.Late
ncy
TTP.AmpS
um
Global.M
ean
Global.MaxDefle
ction
Column 17 - 23
HR hr-sub stress level stress-sub temp temp-sub preference

num ID Event.
Nr
Event.Name CDA.nSC
R
CDA.Laten
cy
CDA.AmpSum CDA.SCR
1 sa1 1 baseline 0

0 0.00644915
3
1 sa1 2 arti lighting 6 1 0.096448173 0.02526754
9
1 sa1 3 daylight 0

0 0.00136564
1
1 sa1 4 highIllu 1 4.25 0.012477743 0.00733287
8
1 sa1 5 lowIllu 0

0 0.00620317
6
1 sa1 6 cityView 0

0 0.00307861
1
1 sa1 7 greenView 2 2 0.038900343 0.01201656
4
1 sa1 8 Plants 1 2 0.010266924 0.00099878
5
1 sa1 9 noPlants 4 1 0.044509713 0.01293183
7
1 sa1 10 bright 1 3.75 0.011035638 0.00900837
9
1 sa1 11 red 0

0 0.00801528
1
1 sa1 12 fabric 3 1 0.044007092 0.01326478
3
1 sa1 13 wood 2 2 0.041222308 0.01771486
9
182

1 sa1 14 highCeiling 2 1.75 0.035640812 0.00750249
1
1 sa1 15 lowCeiling 1 5 0.016686486 0.00705490
3
2 sb1 1 baseline 0

0 0.00058260
1
2 sb1 2 arti lighting 0

0 0.00104809
1
2 sb1 3 daylight 0

0 0.00099343
5
2 sb1 4 highIllu 0

0 0.00057448
2 sb1 5 lowIllu 0

0 0.00159094
4
2 sb1 6 cityView 0

0 0.00102107
6
2 sb1 7 greenView 0

0 0.00115254
8
2 sb1 8 Plants 0

0 0.00135918
6
2 sb1 9 noPlants 0

0 0.00134485
3
2 sb1 10 bright 0

0 0.00137619
2
2 sb1 11 red 0

0 0.00056582
7
2 sb1 12 fabric 0

0 0.00063509
9
2 sb1 13 wood 0

0 0.00059548
2
2 sb1 14 highCeiling 0

0 0.00098802
1
2 sb1 15 lowCeiling 0

0 0.00094629
3 sb2 1 baseline 0

0 0.00134858
3
3 sb2 2 arti lighting 0

0 0.00113959
3
3 sb2 3 daylight 0

0 0.00157332
7
3 sb2 4 highIllu 0

0 0.00201801
9
3 sb2 5 lowIllu 0

0 0
3 sb2 6 cityView 1 1.375 0.016445395 0.00711612
8
3 sb2 7 greenView 1 1.875 0.064551984 0.02908079
6
3 sb2 8 Plants 1 3.375 0.077604815 0.03231447
5
3 sb2 9 noPlants 2 2.375 0.032140687 0.01060956
9
183

3 sb2 10 bright 2 1.875 0.148334391 0.05312558
7
3 sb2 11 red 2 1.375 0.027225812 0.00707043
6
3 sb2 12 fabric 0

0 0.00643580
4
3 sb2 13 wood 1 3.875 0.022980648 0.01021615
3 sb2 14 highCeiling 1 4.875 0.039647677 0.01308604
5
3 sb2 15 lowCeiling 1 2.375 0.016857439 0.01043513
6
4 sb3 1 baseline 0

0 0.00090232
3
4 sb3 2 arti lighting 0

0 0.00229223
9
4 sb3 3 daylight 0

0 0.00099256
9
4 sb3 4 highIllu 0

0 0.00087073
3
4 sb3 5 lowIllu 0

0 0.00081411
5
4 sb3 6 cityView 0

0 0.00067136
3
4 sb3 7 greenView 0

0 0.00182673
4
4 sb3 8 Plants 0

0 0.00499042
5
4 sb3 9 noPlants 0

0 0.00219788
6
4 sb3 10 bright 0

0 0.00101851
3
4 sb3 11 red 0

0 0.00314171
4 sb3 12 fabric 0

0 0.00101287
4
4 sb3 13 wood 0

0 0
4 sb3 14 highCeiling 0

0 0.00087222
7
4 sb3 15 lowCeiling 2 4.5 0.041141443 0.00984419
8
5 sc1 1 baseline 1 4 0.077561345 0.01355209
3
5 sc1 2 arti lighting 1 3.25 0.152878281 0.03272831
3
5 sc1 3 daylight 0

0 0.00672583
3
5 sc1 4 highIllu 4 1.75 0.046058256 0.00287758
5
5 sc1 5 lowIllu 1 2.75 0.148094406 0.03338171
6
184

5 sc1 6 cityView 0

0 0.00183390
4
5 sc1 7 greenView 2 1.75 0.155792441 0.02039272
2
5 sc1 8 Plants 1 4 0.098288418 0.01604647
4
5 sc1 9 noPlants 1 2 0.179884039 0.03690637
4
5 sc1 10 red 0

0 0.03130709
1
5 sc1 11 bright 1 2.75 0.013629437 0.00286959
9
5 sc1 12 wood 0

0 0.04854432
6
5 sc1 13 fabric 1 4.75 0.018022998 0.00208834
5
5 sc1 14 highCeiling 0

0 0.010855
5 sc1 15 lowCeiling 1 1.75 0.040867438 0.00675492
5
6 sd1 1 baseline 4 1.25 0.057802557 0.01093492
7
6 sd1 2 daylight 1 3.5 0.01902222 0.00351772
7
6 sd1 3 arti lighting 3 2 0.045227564 0.00847859
6 sd1 4 lowIllu 2 1.5 0.106943877 0.01730761
2
6 sd1 5 highIllu 3 1.25 0.138675198 0.02596440
9
6 sd1 6 greenView 3 2.25 0.135285998 0.02180387
6 sd1 7 cityView 3 1.75 0.075564525 0.01789543
6 sd1 8 noPlants 1 3.75 0.028720142 0.01155648
1
6 sd1 9 Plants 3 1.75 0.105280069 0.01128713
4
6 sd1 10 bright 1 1.5 0.023680736 0.00684124
8
6 sd1 11 red 2 1.25 0.038739921 0.00620262
1
6 sd1 12 fabric 3 1.75 0.198141873 0.03839913
3
6 sd1 13 wood 2 2.75 0.028232026 0.0066512
6 sd1 14 lowCeiling 2 1.5 0.202232588 0.04018229
1
6 sd1 15 highCeiling 4 2 0.157074318 0.03118833
8
7 se1 1 baseline 2 3 0.026741193 0.00855564
2
7 se1 2 arti lighting 2 1.25 0.413296945 0.06992744
185

7 se1 3 daylight 0

0 0.00906707
8
7 se1 4 highIllu 2 2.25 0.190068229 0.04147415
6
7 se1 5 lowIllu 2 1.5 0.479401359 0.10082392
6
7 se1 6 cityView 2 1.25 0.265913795 0.04560005
3
7 se1 7 greenView 1 3.75 0.322563797 0.07381729
4
7 se1 8 Plants 1 1.5 0.455232877 0.10470469
8
7 se1 9 noPlants 4 2.5 0.05971905 0.00699009
5
7 se1 10 red 2 3 0.539147992 0.12080359
5
7 se1 11 bright 1 4.25 0.011053685 0.00182941
9
7 se1 12 wood 3 1.25 0.706652107 0.13880964
9
7 se1 13 fabric 0

0 0.00315453
1
7 se1 14 highCeiling 2 3.75 0.055436382 0.01176977
9
7 se1 15 lowCeiling 1 2 0.539810154 0.12210757
9
8 se2 1 baseline 0

0 0.00120615
2
8 se2 2 arti lighting 0

0 0.00141464
8
8 se2 3 daylight 0

0 0.00049266
9
8 se2 4 highIllu 0

0 0.00048420
6
8 se2 5 lowIllu 0

0 0.00095201
6
8 se2 6 cityView 0

0 0.00072811
5
8 se2 7 greenView 0

0 0.00056981
4
8 se2 8 Plants 0

0 0.00102792
7
8 se2 9 noPlants 0

0 0.00057400
7
8 se2 10 red 0

0 0.00103348
4
8 se2 11 bright 0

0 0.00045689
1
8 se2 12 wood 0

0 0.00074783
186

8 se2 13 fabric 0

0 0.00072998
7
8 se2 14 highCeiling 0

0 0.00039169
3
8 se2 15 lowCeiling 0

0 0.00063815
8
9 se3 1 baseline 0

0 0.00051119
9 se3 2 arti lighting 0

0 0.00015092
4
9 se3 3 daylight 0

0 0.00159158
2
9 se3 4 highIllu 0

0 0.00072247
9 se3 5 lowIllu 7 1 0.226158241 0.04878792
2
9 se3 6 cityView 0

0 0.00111271
4
9 se3 7 greenView 0

0 4.56E-05
9 se3 8 Plants 0

0 0.00249567
6
9 se3 9 noPlants 0

0 0.00418072
6
9 se3 10 red 2 3.5 0.147098698 0.04014252
7
9 se3 11 bright 1 2.5 0.181027605 0.05117458
4
9 se3 12 wood 0

0 0.00048652
6
9 se3 13 fabric 0

0 0.00183885
3
9 se3 14 highCeiling 1 1.5 0.261695267 0.06454435
5
9 se3 15 LowCeilingMissi
ng
0

0 0.00132440
9
10 se4 1 baseline 0

0 0.00071882
6
10 se4 2 arti lighting 0

0 0.00104012
3
10 se4 3 daylight 0

0 0.00042231
9
10 se4 4 highIllu 0

0 0.00212031
2
10 se4 5 lowIllu 1 2 0.011515422 0.00463195
2
10 se4 6 cityView 0

0 0.00404427
9
10 se4 7 greenView 0

0 0.00222298
6
10 se4 8 Plants 0

0 0.00029949
5
187

10 se4 9 noPlants 1 4.75 0.030000507 0.00716025
9
10 se4 10 red 3 3.25 0.067149837 0.01239510
5
10 se4 11 bright 1 1.25 0.014508835 0.00576213
1
10 se4 12 wood 0

0 0
10 se4 13 fabric 0

0 0.00226980
6
10 se4 14 highCeiling 1 2.25 0.064852674 0.02664364
3
10 se4 15 lowCeiling 1 2.5 0.138798925 0.03381577
1
11 se5 1 baseline 1 3.5 0.010716403 0.00878005
1
11 se5 2 arti lighting 3 1.5 0.055283678 0.01516815
5
11 se5 3 daylight 2 2.25 0.042496648 0.01286419
7
11 se5 4 highIllu 3 1.25 0.244112428 0.04786226
9
11 se5 5 lowIllu 1 1.25 0.012273708 0.00913616
7
11 se5 6 cityView 1 1.25 0.015744049 0.00926729
4
11 se5 7 greenView 1 3 0.091385504 0.02123401
7
11 se5 8 Plants 0

0 0.00069582
8
11 se5 9 noPlants 1 5 0.11060931 0.01542390
6
11 se5 10 red 0

0 0.01495873
11 se5 11 bright 1 4 0.320509778 0.07083883
4
11 se5 12 wood 1 3.5 0.059289198 0.01596744
3
11 se5 13 fabric 0

0 0
11 se5 14 highCeiling 2 4 0.187051029 0.02509688
6
11 se5 15 lowCeiling 1 3.75 0.026597648 0.00567269
1
12 se6 1 baseline 0

0 0.00076062
1
12 se6 2 daylight 0

0 0.00145866
2
12 se6 3 arti lighting 0

0 0.00191356
1
12 se6 4 lowIllu 1 1.75 0.011726066 0.00224476
3
188

12 se6 5 highIllu 2 2.5 0.026342222 0.00328291
8
12 se6 6 greenView 3 1.5 0.058336696 0.00516642
3
12 se6 7 cityView 0

0 0.00169889
12 se6 8 noPlants 0

0 0.00125701
1
12 se6 9 Plants 0

0 0.00119678
12 se6 10 bright 0

0 0.00217234
5
12 se6 11 red 0

0 0.00253746
6
12 se6 12 fabric 0

0 0.00107069
4
12 se6 13 wood 0

0 0.00145363
4
12 se6 14 lowCeiling 0

0 0.00132256
6
12 se6 15 highCeiling 4 1 0.055647801 0.00352388
13 sf1 1 baseline 0

0 0.00120233
1
13 sf1 2 arti lighting 0

0 0.00100632
1
13 sf1 3 daylight 0

0 0.00072555
4
13 sf1 4 lowIllu 0

0 0.00117245
6
13 sf1 5 highIllu 0

0 0.00084716
8
13 sf1 6 greenView 0

0 0.00035262
3
13 sf1 7 cityView 0

0 0.00150087
9
13 sf1 8 noPlants 0

0 0.00055248
2
13 sf1 9 Plants 0

0 0.00214862
6
13 sf1 10 bright 0

0 0.00069285
8
13 sf1 11 red 1 1.25 0.025254371 0.00488294
2
13 sf1 12 fabric 2 1.25 0.075657114 0.01692712
3
13 sf1 13 wood 1 3.5 0.018486841 0.00793547
3
13 sf1 14 lowCeiling 0

0 0.00086377
8
13 sf1 15 highCeiling 2 1.5 0.076744208 0.01738243
8
189

14 sf2 1 baseline 0

0 0.00028129
1
14 sf2 2 daylight 0

0 0.00063077
1
14 sf2 3 arti lighting 0

0 0.00074321
5
14 sf2 4 lowIllu 0

0 0.00072491
4
14 sf2 5 highIllu 0

0 0.00101617
4
14 sf2 6 cityView 0

0 0.00064187
4
14 sf2 7 greenView 0

0 0.00095528
7
14 sf2 8 noPlants 0

0 0.00068556
6
14 sf2 9 Plants 0

0 0.00100318
7
14 sf2 10 bright 0

0 0.00120805
8
14 sf2 11 red 0

0 0.00076741
14 sf2 12 fabric 0

0 0.00082253
7
14 sf2 13 wood 0

0 0.00114534
14 sf2 14 lowCeiling 0

0 0.00041677
3
14 sf2 15 highCeiling 0

0 0.00167042
7
15 sf3 1 baseline 0

0 0.00139529
6
15 sf3 2 arti lighting 0

0 0.00096612
9
15 sf3 3 daylight 0

0 0.00071714
9
15 sf3 4 highIllu 0

0 0.00081355
15 sf3 5 lowIllu 0

0 0.00045393
7
15 sf3 6 cityView 0

0 0.00225158
5
15 sf3 7 greenView 0

0 0.00088195
5
15 sf3 8 Plants 0

0 0.00167848
9
15 sf3 9 noPlants 0

0 0.00179878
15 sf3 10 red 0

0 0.00144870
5
15 sf3 11 bright 0

0 0.00235413
15 sf3 12 wood 0

0 0.00109454
5
190

15 sf3 13 fabric 0

0 0.00106388
8
15 sf3 14 highCeiling 0

0 0.00064591
15 sf3 15 lowCeiling 0

0 0.00059738
16 sf4 1 baseline 0

0 0.00129315
3
16 sf4 2 daylight 0

0 0.00074974
7
16 sf4 3 arti lighting 0

0 0.00086520
6
16 sf4 4 lowIllu 0

0 0.00064758
4
16 sf4 5 highIllu 0

0 0.00065626
6
16 sf4 6 greenView 0

0 0.00071690
3
16 sf4 7 cityView 0

0 0.00077046
6
16 sf4 8 noPlants 0

0 0.00092279
9
16 sf4 9 Plants 0

0 0.00092098
1
16 sf4 10 bright 0

0 0.00073081
2
16 sf4 11 red 0

0 0.00080976
3
16 sf4 12 fabric 0

0 0.00082405
3
16 sf4 13 wood 0

0 0.00054590
8
16 sf4 14 lowCeiling 0

0 0.00097458
5
16 sf4 15 highCeiling 0

0 0.00129529
4
17 sf5 1 baseline 0

0 0.00046451
4
17 sf5 2 arti lighting 0

0 0.00097844
3
17 sf5 3 daylight 0

0 0.00160612
6
17 sf5 4 lowIllu 0

0 0.00127856
7
17 sf5 5 highIllu 0

0 0.00102496
6
17 sf5 6 greenView 0

0 0.00154559
5
17 sf5 7 cityView 0

0 0.00546229
1
17 sf5 8 noPlants 0

0 0.00419321
6
191

17 sf5 9 Plants 0

0 0.00215787
7
17 sf5 10 bright 0

0 0.00142648
3
17 sf5 11 red 0

0 0.00088983
17 sf5 12 fabric 0

0 0.00247519
7
17 sf5 13 wood 3 1.75 0.063122598 0.01487433
17 sf5 14 lowCeiling 0

0 0.00297734
6
17 sf5 15 highCeiling 1 1 0.012086636 0.00274140
4
18 sg1 1 baseline 0

0 0.00073378
2
18 sg1 2 daylight 0

0 0.00083964
5
18 sg1 3 arti lighting 0

0 0.00077751
7
18 sg1 4 lowIllu 0

0 0.00083093
7
18 sg1 5 highIllu 0

0 0.00126348
7
18 sg1 6 greenView 0

0 0.00104541
6
18 sg1 7 cityView 0

0 0.00076153
8
18 sg1 8 noPlants 0

0 0.00103902
3
18 sg1 9 Plants 0

0 0.00119999
1
18 sg1 10 bright 0

0 0.00083381
7
18 sg1 11 red 0

0 0.00079127
6
18 sg1 12 fabric 0

0 0.00081410
2
18 sg1 13 wood 0

0 0.00053659
7
18 sg1 14 lowCeiling 0

0 0.00113393
4
18 sg1 15 highCeiling 0

0 0.00074061
5
19 sh1 1 baseline 0

0 0.00131951
1
19 sh1 2 arti lighting 0

0 0.00236099
1
19 sh1 3 daylight 0

0 0.00088100
7
19 sh1 4 lowIllu 0

0 0.00170917
5
192

19 sh1 5 highIllu 0

0 0.00055480
7
19 sh1 6 cityView 0

0 0.00129825
4
19 sh1 7 greenView 0

0 0.00040575
8
19 sh1 8 noPlants 0

0 0.00107032
4
19 sh1 9 Plants 0

0 0.00036237
3
19 sh1 10 red 0

0 0.00058650
9
19 sh1 11 bright 0

0 0.00172627
2
19 sh1 12 wood 0

0 0.00086444
1
19 sh1 13 fabric 0

0 0.00248827
4
19 sh1 14 highCeiling 0

0 0.00300646
4
19 sh1 15 lowCeiling 1 1 0.0137726 0.00604261
8
20 sh2 1 baseline 0

0 0.00108526
2
20 sh2 2 arti lighting 0

0 0.00064559
2
20 sh2 3 daylight 0

0 0.00026784
6
20 sh2 4 highIllu 0

0 0.00078515
1
20 sh2 5 lowIllu 0

0 0.00092220
5
20 sh2 6 cityView 0

0 0.00057158
6
20 sh2 7 greenView 0

0 0.00076739
6
20 sh2 8 Plants 0

0 0.00111488
9
20 sh2 9 noPlants 0

0 0.00084110
2
20 sh2 10 red 0

0 0.00087327
6
20 sh2 11 bright 0

0 0.00089975
8
20 sh2 12 wood 0

0 0.00106749
1
20 sh2 13 fabric 0

0 0.00055619
4
20 sh2 14 highCeiling 0

0 0.00104092
2
193

20 sh2 15 lowCeiling 0

0 0.00066070
4
21 sh3 1 baseline 0

0 0.00099028
1
21 sh3 2 arti lighting 0

0 0.00163299
21 sh3 3 daylight 2 1.25 0.022635806 0.00391757
3
21 sh3 4 highIllu 2 2.25 0.030130847 0.00857632
8
21 sh3 5 lowIllu 2 2.25 0.041402481 0.01004420
6
21 sh3 6 cityView 0

0 0.00375991
3
21 sh3 7 greenView 0

0 0.00143609
5
21 sh3 8 Plants 2 1 0.046716669 0.00925407
5
21 sh3 9 noPlants 2 2.25 0.047104606 0.01015617
8
21 sh3 10 red 0

0 0.00140093
5
21 sh3 11 bright 0

0 0.00564629
6
21 sh3 12 wood 0

0 0.00265095
9
21 sh3 13 fabric 2 1.5 0.028302498 0.004651
21 sh3 14 highCeiling 2 1 0.092959906 0.02080611
3
21 sh3 15 lowCeiling 0

0 0.00598052
5
22 sh4 1 baseline 0

0 0.00191779
3
22 sh4 2 daylight 0

0 0.00059110
3
22 sh4 3 arti lighting 1 3.75 0.015657089 0.00374185
5
22 sh4 4 lowIllu 1 3.5 0.046998805 0.01336776
7
22 sh4 5 highIllu 0

0 0.01255500
8
22 sh4 6 greenView 1 4.5 0.020387755 0.00186089
8
22 sh4 7 cityView 4 1 0.091319121 0.01477769
2
22 sh4 8 noPlants 2 1.5 0.12707831 0.02363493
6
22 sh4 9 Plants 2 2 0.021263586 0.00568263
5
22 sh4 10 bright 0

0 0.00442289
8
194

22 sh4 11 red 1 5 0.019446392 0.00497550
8
22 sh4 12 fabric 3 1.25 0.078717949 0.01167772
8
22 sh4 13 wood 2 3.25 0.041339698 0.00690641
7
22 sh4 14 lowCeiling 2 1.75 0.147485499 0.03042544
1
22 sh4 15 highCeiling 2 1.25 0.073710323 0.01534824
9
23 sh5 1 baseline 0

0 0.00072406
23 sh5 2 daylight 0

0 0.00098272
8
23 sh5 3 arti lighting 0

0 0.00103186
5
23 sh5 4 lowIllu 0

0 0.00094133
9
23 sh5 5 highIllu 0

0 0.00072690
7
23 sh5 6 greenView 0

0 0.00076034
23 sh5 7 cityView 0

0 0.00106250
2
23 sh5 8 noPlants 0

0 0.00084880
9
23 sh5 9 Plants 0

0 0.00136023
8
23 sh5 10 bright 0

0 0.00081949
9
23 sh5 11 red 0

0 0.00087702
9
23 sh5 12 fabric 0

0 0.00089586
6
23 sh5 13 wood 0

0 0.00121694
1
23 sh5 14 lowCeiling 0

0 0.00098171
23 sh5 15 highCeiling 0

0 0.00087440
2
24 si1 1 baseline 0

0 0.00121043
1
24 si1 2 daylight 0

0 0.00068463
2
24 si1 3 arti lighting 0

0 0.00099099
3
24 si1 4 lowIllu 0

0 0.00071427
7
24 si1 5 highIllu 0

0 0.00105672
8
24 si1 6 greenView 0

0 0.00065934
9
24 si1 7 cityView 0

0 0.00080212
195

24 si1 8 noPlants 0

0 0.00059400
2
24 si1 9 Plants 0

0 0.00074711
8
24 si1 10 red 0

0 0.00124968
24 si1 11 bright 0

0 0.00057179
7
24 si1 12 fabric 0

0 0.00078580
1
24 si1 13 wood 0

0 0.00046811
9
24 si1 14 lowCeiling 1 4.25 0.024220414 0.00236944
7
24 si1 15 highCeiling 0

0 0.00110611
4
25 si2 1 baseline 0

0 0.00295077
25 si2 2 daylight 0

0 0
25 si2 3 arti lighting 1 3.5 0.163798875 0.03416526
2
25 si2 4 highIllu 2 3.25 0.148427198 0.02913328
2
25 si2 5 lowIllu 1 1.75 0.068381317 0.01002364
2
25 si2 6 cityView 1 2.75 0.05309453 0.01219985
3
25 si2 7 greenView 2 1 0.481776413 0.09410121
9
25 si2 8 noPlants 3 2.25 0.28196431 0.05595260
4
25 si2 9 Plants 3 1.25 1.891334959 0.33652758
4
25 si2 10 bright 2 1.25 0.517901609 0.03945036
3
25 si2 11 red 3 1 1.389825121 0.21755426
2
25 si2 12 wood 3 2 1.423312757 0.28170787
1
25 si2 13 fabric 2 1 0.204025045 0.04440032
2
25 si2 14 highCeiling 3 1.5 0.362366055 0.06213514
4
25 si2 15 lowCeiling 1 3.5 0.13395453 0.02845540
2
26 si3 1 baseline 0

0 0.00088805
6
26 si3 2 arti lighting 0

0 0.00045362
26 si3 3 daylight 0

0 0.00047895
9
26 si3 4 lowIllu 0

0 0.00104057
9
196

26 si3 5 highIllu 0

0 0.00102821
9
26 si3 6 cityView 0

0 0.00178717
7
26 si3 7 greenView 0

0 0.00065889
7
26 si3 8 Plants 0

0 0.00082732
8
26 si3 9 noPlants 0

0 0.00102328
9
26 si3 10 bright 0

0 0.00099254
9
26 si3 11 red 0

0 0.00173272
8
26 si3 12 wood 1 4 0.031841873 0.00668043
7
26 si3 13 fabric 2 1.25 0.060522755 0.00458119
8
26 si3 14 lowCeiling 0

0 0.00228514
26 si3 15 highCeiling 0

0 0.00320868
3
27 sj1 1 baseline 1 1.75 0.034104053 0.00560891
5
27 sj1 2 arti lighting 1 2.75 0.025027242 0.00655232
4
27 sj1 3 daylight 1 4.75 0.025515716 0.00402173
27 sj1 4 highIllu 2 2.25 0.094020903 0.01491608
5
27 sj1 5 lowIllu 2 1.5 0.053059186 0.00918867
4
27 sj1 6 cityView 2 3.25 0.075405623 0.01372385
9
27 sj1 7 greenView 1 4.75 0.014198612 0.00529249
4
27 sj1 8 Plants 1 3 0.103160341 0.01915248
8
27 sj1 9 noPlants 1 4.5 0.086098951 0.01750011
1
27 sj1 10 red 1 3.5 0.063467958 0.01186666
1
27 sj1 11 bright 3 1.25 0.084431116 0.01238587
1
27 sj1 12 wood 0

0 0.00101633
7
27 sj1 13 fabric 2 1.5 0.054908069 0.01110325
8
27 sj1 14 highCeiling 1 4.5 0.107299316 0.01491102
2
27 sj1 15 lowCeiling 2 1.5 0.116427247 0.02130468
4
197

28 sk1 1 baseline 0

0 0.00082314
1
28 sk1 2 daylight 0

0 0.00057234
9
28 sk1 3 arti lighting 0

0 0.00141188
5
28 sk1 4 highIllu 0

0 0.00033154
8
28 sk1 5 lowIllu 0

0 0.00086526
2
28 sk1 6 cityView 0

0 0.00158721
1
28 sk1 7 greenView 0

0 0.00047207
1
28 sk1 8 Plants 0

0 0.00121515
2
28 sk1 9 noPlants 0

0 0.00080965
1
28 sk1 10 bright 0

0 0.00184324
9
28 sk1 11 red 0

0 0.00039879
7
28 sk1 12 fabric 0

0 0.00136189
7
28 sk1 13 wood 0

0 0.00144849
6
28 sk1 14 lowCeiling 1 5 0.017923167 0.00399046
9
28 sk1 15 highCeiling 1 3.25 0.088331239 0.02531281
29 sk2 1 baseline 3 1 0.088725211 0.01254400
6
29 sk2 2 daylight 3 3 0.055768185 0.00940258
7
29 sk2 3 arti lighting 4 2.25 0.235512282 0.04818177
2
29 sk2 4 lowIllu 3 1.25 0.280625927 0.05508443
3
29 sk2 5 highIllu 5 1.25 0.508426004 0.09455180
7
29 sk2 6 greenView 4 1.25 0.967441213 0.18413148
1
29 sk2 7 cityView 6 1.5 1.094518789 0.17708658
29 sk2 8 Plants 3 2 0.230500006 0.04595012
8
29 sk2 9 noPlants 3 1.5 0.402257073 0.07814461
29 sk2 10 bright 5 2 0.501281678 0.09650264
7
29 sk2 11 red 4 1.25 0.290947029 0.05638942
29 sk2 12 wood 2 2.25 0.237861757 0.05768980
9
198

29 sk2 13 fabric 3 1.25 0.387254292 0.07381291
4
29 sk2 14 lowCeiling 4 2.25 0.358390056 0.06877455
29 sk2 15 highCeiling 4 2 0.482153832 0.08605841
4
30 sk3 1 baseline 0

0 0.00051426
7
30 sk3 2 arti lighting 0

0 0.00098218
9
30 sk3 3 daylight 0

0 0.00116780
9
30 sk3 4 highIllu 0

0 0.00052570
5
30 sk3 5 lowIllu 0

0 0.00098901
6
30 sk3 6 cityView 1 4 0.018468309 0.00939156
3
30 sk3 7 greenView 1 2.75 0.011220071 0.00328091
2
30 sk3 8 Plants 1 3 0.029389068 0.01403522
7
30 sk3 9 noPlants 1 1.5 0.121771588 0.02805719
6
30 sk3 10 red 1 1.25 0.013257528 0.00230640
9
30 sk3 11 bright 0

0 0
30 sk3 12 wood 3 1.25 0.172177246 0.04112247
30 sk3 13 fabric 2 1.5 0.105516209 0.02459990
8
30 sk3 14 highCeiling 2 1 0.337087009 0.04818342
3
30 sk3 15 lowCeiling 1 2.5 0.010778153 0.00659130
6

CDA.ISC
R
CDA.Phasic
Max
CDA.To
nic
TTP.nS
CR
TTP.Late
ncy
TTP.Amp
Sum
Global.M
ean
Global.MaxDefl
ection
0.103186
453
0.126894 0.357776 0

0 0.399609 0.005661
0.404280
786
0.151989 0.384785 0

0 0.461636
5
0.006662
0.021850
249
0.056798 0.51392 1 5 0.025301 0.523859
8
0.006369
0.117326
055
0.1151 0.528872 1 4 0.015912 0.543737
2
0.013909
0.099250
811
0.136959 0.533671 0

0 0.546228 0.012128
0.049257
783
0.090904 0.561368 0

0 0.569669
2
0.005879
199

0.192265
017
0.201235 0.548114 0

0 0.570158
2
0.025273
0.015980
568
0.066846 0.522065 0

0 0.529863
2
0.004948
0.206909
386
0.095431 0.444167 0

0 0.466847
1
0.016797
0.144134
066
0.20049 0.389522 0

0 0.410388
1
0.00685
0.128244
489
0.196834 0.492608 0

0 0.515163
5
0.007565
0.212236
523
0.19214 0.471223 1 2 0.022605 0.506249
3
0.024224
0.283437
902
0.661947 0.418326 3 1.75 0.074134 0.451134
1
0.03433
0.120039
859
0.364245 0.440333 2 1.5 0.027158 0.464507
2
0.019288
0.112878
442
0.855281 0.388375 2 2 0.054316 0.421989
8
0.046102
0.004660
81
0.001826 0.00945 0

0 0.010618
9
0.000965
0.008384
73
0.003392 0.010274 0

0 0.012150
4
0.001429
0.007947
483
0.003292 0.013058 0

0 0.015104
6
0.001368
0.004595
839
0.002354 0.013275 0

0 0.014380
3
0.000997
0.012727
554
0.00392 0.012284 0

0 0.015435
3
0.001342
0.008168
606
0.002849 0.014889 0

0 0.016877
3
0.001001
0.009220
381
0.002778 0.014562 0

0 0.016975
8
0.0011
0.010873
484
0.004898 0.015736 0

0 0.018446
1
0.001882
0.010758
825
0.00308 0.017592 0

0 0.020211
1
0.000608
0.011009
536
0.003237 0.01822 0

0 0.020987
4
0.000524
0.004526
617
0.00206 0.022134 0

0 0.023230
2
0.001534
0.005080
792
0.00253 0.021588 0

0 0.022945
3
0.002327
0.004763
858
0.002284 0.021999 0

0 0.023209
4
0.001205
0.007904
165
0.003163 0.022462 0

0 0.024441 0.002593
0.007570
316
0.003251 0.026296 0

0 0.028215
5
0.002157
0.010788
661
0.009748 0.114449 0

0 0.115992 0.00204
200

0.009116
743
0.005993 0.155606 0

0 0.157828
4
0.001914
0.012586
618
0.010873 0.194336 0

0 0.197138
5
0.004124
0.016144
155
0.007696 0.261414 0

0 0.266485
4
0.003421
0 0.004239 0.561561 0

0 0.562877
8
0.00184
0.056929
022
0.032091 0.775292 0

0 0.792733
3
0.002168
0.232646
365
0.113125 0.827248 0

0 0.8825 0.034536
0.258515
798
0.161837 1.104594 1 1.875 0.084646 1.147176
8
0.084646
0.084876
555
0.037805 1.151782 0

0 1.169085
3
0.007912
0.425004
696
0.173851 1.27941 1 2.875 0.074669 1.348843
6
0.098014
0.056563
489
0.035785 1.305739 1 4.875 0.011794 1.327002
3
0.001082
0.051486
434
0.041044 1.318896 0

0 1.334278
6
0.001984
0.081729
199
0.047069 1.322651 1 3.375 0.011549 1.341516
8
0.011103
0.104688
356
0.058601 1.251663 1 1.875 0.028236 1.270102
1
0.021657
0.083481
086
0.040288 1.125589 0

0 1.156468
7
0
0.014437
163
0.028152 0.260438 0

0 0.262865
5
0.005
0.036675
82
0.045593 0.270191 0

0 0.274506
3
0.007166
0.015881
096
0.041575 0.285049 0

0 0.288777
5
0.004805
0.013931
734
0.034467 0.303597 0

0 0.307328
8
0.003169
0.013025
848
0.03955 0.315736 0

0 0.318854
6
0.003342
0.010741
802
0.04229 0.324517 0

0 0.328049
2
0.004039
0.029227
747
0.033392 0.360134 0

0 0.365288
7
0.008725
0.079846
802
0.063374 0.457227 1 1.5 0.014968 0.470694
7
0.015724
0.035166
174
0.030381 0.562106 0

0 0.577942
6
0.00173
0.016296
202
0.073292 0.622853 1 4.75 0.045142 0.625449
9
0.002374
0.050267
359
0.041022 0.628335 0

0 0.648265
8
0.002071
201

0.016205
988
0.030365 0.631599 0

0 0.639047
4
0.002197
0 0.021032 0.637071 0

0 0.647341
5
0.001202
0.013955
625
0.037344 0.633724 0

0 0.659879
1
0.002008
0.157507
166
0.151213 0.650913 1 3.5 0.043746 0.661912
6
0.034124
0.216833
48
0.077682 1.059529 0

0 1.111065
1
0.089167
0.523653
003
0.15329 1.299021 0

0 1.419340
1
0.079536
0.107613
329
0.045274 1.330544 0

0 1.356628
2
0.002715
0.046041
354
0.013247 1.262833 0

0 1.272564
8
0.001451
0.534107
459
0.148582 1.560776 1 1.5 0.033113 1.684804
6
0.034216
0.029342
457
0.009798 1.420517 0

0 1.427476
2
0.003405
0.326283
559
0.112029 1.480497 0

0 1.557319
8
0.016261
0.256743
585
0.098407 1.377399 0

0 1.437871
6
0.105909
0.590501
981
0.179947 1.53343 0

0 1.67181 0.020323
0.500913
45
0.232082 1.688792 0

0 1.808805
1
0
0.045913
59
0.015938 1.90629 0

0 1.916703
1
0.004819
0.776709
217
0.258965 1.954008 0

0 2.139545
8
0.000191
0.033413
518
0.018062 2.085395 1 3.25 0.012964 2.092751
9
0.014082
0.173679
993
0.063117 2.049021 0

0 2.089408
2
0
0.108078
8
0.040962 2.042233 0

0 2.068558
4
0.009051
0.174958
827
0.082607 0.668552 1 2.5 0.024657 0.719674
1
0.025596
0.056283
632
0.12102 0.801122 1 3 0.024271 0.823444
8
0.025859
0.135657
434
0.076108 1.058338 2 1 0.034688 1.076991
4
0.025774
0.276921
788
0.23814 1.132179 1 5 0.015785 1.207244
9
0.066921
0.415430
549
0.186738 1.232547 1 3.25 0.028909 1.359522
9
0.029891
0.348861
914
0.29453 1.40488 3 1.5 0.115353 1.484041
8
0.079241
202

0.286326
874
0.271699 1.580884 2 1.5 0.035095 1.666880
4
0.024836
0.184903
702
0.145021 1.777552 1 3.5 0.014449 1.857353
8
0.018015
0.180594
149
0.261211 1.591618 1 4.75 0.063452 1.775669
5
0.007484
0.109459
972
0.155707 1.830103 0

0 1.862754
2
0.020734
0.099241
928
0.098177 1.877858 1 2 0.018896 1.901850
5
0.023076
0.614386
121
0.444519 1.911283 2 1.25 0.100784 2.063416
1
0.096686
0.106419
203
0.214126 2.239142 0

0 2.389893 0.010499
0.642916
654
0.521997 2.346592 0

0 2.464276
2
0.164567
0.499013
409
0.618737 2.380452 3 1.5 0.088522 2.572036
5
0.049899
0.136890
272
0.078541 4.382122 0

0 4.420854
6
0.001744
1.118839
032
1.098456 4.359727 0

0 4.612628
7
0.323285
0.145073
249
0.154562 4.149201 0

0 4.259010
5
0.00386
0.663586
5
0.549946 3.944001 1 1.5 0.120915 4.082687
9
0.12194
1.613182
809
1.361101 4.064586 0

0 4.388998
7
0.459567
0.729600
844
0.857072 3.90374 0

0 4.090853
5
0.165971
1.181076
711
0.848762 3.698325 1 2.75 0.324815 3.788014 0.321636
1.675275
17
1.304448 3.55714 1 5 0.314217 3.903596
1
0.392962
0.111841
527
0.084769 3.585589 0

0 3.734971
5
0
1.932857
516
1.376663 3.381335 1 1.75 0.514771 3.645830
5
0.518155
0.029270
697
0.124504 3.296867 1 4 0.011587 3.301909
9
0.013413
2.220954
382
0.833198 3.003747 1 3.25 0.141144 3.326170
8
0.412262
0.050472
493
0.044412 3.080745 0

0 3.091033
2
0.002043
0.188316
458
0.151765 2.908524 1 3 0.04765 2.926362
4
0.04156
1.953721
269
1.699085 2.965182 0

0 3.307669
9
0.556827
0.019298
431
0.005159 0.002232 0

0 0.006776
4
0.00088
203

0.022634
373
0.0067 0.00102 0

0 0.006331
6
0.003624
0.007882
702
0.002483 0.004414 0

0 0.006215
1
0.001843
0.007747
3
0.002766 0.006374 0

0 0.008180
6
0.002204
0.015232
249
0.004212 0.005456 0

0 0.009041
4
0.001566
0.011649
841
0.003535 0.005 0

0 0.007697
1
0.001933
0.009117
02
0.003444 0.005826 0

0 0.007966
5
0.002167
0.016446
84
0.004435 0.00591 0

0 0.009831
8
0.002442
0.009184
111
0.003062 0.006411 0

0 0.008537
5
0.002876
0.016535
744
0.004785 0.006911 0

0 0.010710
3
0.002802
0.007310
264
0.003291 0.00942 0

0 0.011163
5
0.002205
0.011965
284
0.003485 0.007191 0

0 0.010022
9
0.001792
0.011679
785
0.003991 0.007343 0

0 0.010086
4
0.001543
0.006267
085
0.002629 0.009672 0

0 0.011159
8
0.003282
0.010210
53
0.003138 0.010036 0

0 0.012446
2
0.002396
0.008179
045
0.04499 1.103957 0

0 1.106703
9
0.002675
0.002414
778
0.024656 1.108615 0

0 1.110585
3
0.002343
0.025465
319
0.029613 1.10861 0

0 1.114109 0.002289
0.011559
521
0.046689 1.130024 0

0 1.134196
3
0.002404
0.780606
75
0.229255 0.989733 0

0 1.176917
3
0.003081
0.017803
419
0.018528 1.165716 0

0 1.167906
8
0.001038
0.000729
946
0.033115 1.158831 0

0 1.165249
2
0.001226
0.039930
821
0.072784 1.284417 0

0 1.290998
6
0.003392
0.066891
61
0.072123 1.382626 0

0 1.389305
2
0.007816
0.642280
43
0.644252 1.547649 1 2.75 0.41214 1.598738
2
0.157872
0.818793
338
0.624285 1.623082 1 1.25 0.177588 1.725482
2
0.178585
204

0.007784
418
0.028924 1.757522 0

0 1.763796
3
0.000355
0.029421
653
0.064226 1.682483 0

0 1.703865
7
0.003654
1.032709
688
1.225165 1.829354 0

0 2.054944
9
0.217075
0.021190
551
0.036697 2.02749 0

0 2.051865
3

0.011501
21
0.034935 0.042719 0

0 0.045575
4
0.002488
0.016641
972
0.043974 0.052203 0

0 0.055131
6
0.004495
0.006757
1
0.038781 0.065472 0

0 0.073348
9
0.00286
0.033924
989
0.027111 0.073297 0

0 0.078059
7
0.004504
0.074111
228
0.137781 0.173201 1 1.25 0.012729 0.182303
5
0.017473
0.064708
467
0.049615 0.208199 1 3.5 0.020265 0.220596
7
0.013383
0.035567
777
0.040983 0.273866 0

0 0.282054
6
0.002425
0.004791
919
0.024342 0.28279 1 4.75 0.058043 0.291238
5
0.003496
0.114564
142
0.173488 0.46199 1 3.75 0.039959 0.470509 0.02529
0.198321
688
0.125879 0.508633 1 2.75 0.047417 0.536319
4
0.044311
0.092194
097
0.07098 0.59581 0

0 0.630455
7
0.011343
0 0.037045 0.65947 0

0 0.665714
1
0.001697
0.036316
89
0.049869 0.591171 0

0 0.603985
3
0.005831
0.426298
282
0.349354 0.572184 0

0 0.686893
6
0.080743
0.541052
334
0.3674 0.738876 1 1 0.141094 0.825332
4
0.141717
0.140480
814
0.055692 0.990562 0

0 1.04648 0.000272
0.242690
475
0.273371 1.183174 2 3.5 0.06574 1.267723
5
0.030207
0.205827
149
0.153487 1.08983 0

0 1.166167
8
0.0068
0.765796
302
0.319027 1.107872 1 3.25 0.08036 1.215602
5
0.152623
0.146178
669
0.074422 1.577552 0

0 1.643602
4
0.004573
0.148276
702
0.103263 1.518539 0

0 1.601482
9
0.00136
205

0.339744
279
0.352729 1.768948 1 2.25 0.097087 1.814851
7
0.098623
0.011133
245
0.086658 1.67488 0

0 1.682237
1
0.004369
0.246782
491
0.393836 1.669692 1 4.25 0.11761 1.685000
4
0.044477
0.239339
678
0.439207 1.878205 1 4.5 0.266306 1.967214
5
0.010678
1.133421
35
0.883835 1.923478 1 3 0.303408 2.029823
3
0.280428
0.255479
084
0.273096 2.004048 1 2.25 0.081638 2.028033
6
0.059429
0 0.021302 1.893498 0

0 1.928538
5
0
0.401550
177
0.748419 2.04631 1 3.5 0.177618 2.058361
3
0.070527
0.090763
051
0.191586 2.071207 1 3.25 0.021275 2.090610
4
0.022536
0.012169
937
0.005034 1.243401 0

0 1.246348
9
0.00255
0.023338
585
0.009845 1.254759 0

0 1.260170
8
0.005532
0.030616
983
0.010242 1.255448 0

0 1.263053 0.003824
0.035916
201
0.016449 1.282591 0

0 1.291805
3
0.007947
0.052526
693
0.019633 1.290565 0

0 1.302966
1
0.009301
0.082662
761
0.034163 1.286834 1 2.5 0.011101 1.305504
6
0.013009
0.027182
237
0.011656 1.290751 0

0 1.296614
5
0.005384
0.020112
181
0.007915 1.302564 0

0 1.307426
2
0.004313
0.019148
486
0.007351 1.31715 0

0 1.321822
1
0.004198
0.034757
524
0.011992 1.318469 0

0 1.326664
2
0.009723
0.040599
452
0.011792 1.313986 0

0 1.323503 0.003824
0.017131
104
0.008551 1.331762 0

0 1.335464
6
0.005573
0.023258
143
0.011237 1.32656 0

0 1.332269
3
0.006781
0.021161
058
0.009862 1.326491 0

0 1.331400
8
0.005278
0.056382
074
0.023371 1.351977 0

0 1.365878
8
0.009767
0.019237
295
0.024936 0.001947 0

0 0.004734
3
0.003448
206

0.016101
144
0.023102 0.002322 0

0 0.006118
9
0.003631
0.011608
856
0.025439 0.002612 0

0 0.005775
5
0.004264
0.018759
291
0.019996 0.003684 0

0 0.007713
5
0.001082
0.013554
692
0.02465 0.006015 0

0 0.009961
2
0.0033
0.005641
974
0.026977 0.005527 0

0 0.008717
4
0.002733
0.024014
068
0.024399 0.004025 0

0 0.008489
9
0.002472
0.008839
719
0.019837 0.014605 0

0 0.016487
6
0.002363
0.034378
013
0.033563 0.162159 0

0 0.170513
1
0.006102
0.011085
731
0.045107 0.363323 0

0 0.369446
1
0.001868
0.078127
076
0.087952 0.6003 0

0 0.620881
7
0.016246
0.270833
971
0.399938 0.870803 2 2.25 0.045852 0.945675
8
0.051902
0.126967
574
0.09632 1.275161 1 2.25 0.035249 1.328469
7
0.038901
0.013820
453
0.102143 1.944113 0

0 1.967521
3
0.008791
0.278119
007
0.358198 2.147111 1 2.5 0.016289 2.198537
8
0.055036
0.004500
652
0.002443 0.01121 0

0 0.012253
4
0.001683
0.010092
328
0.003431 0.010964 0

0 0.013386
9
0.002201
0.011891
435
0.004016 0.012154 0

0 0.014927
2
0.002498
0.011598
622
0.0044 0.012443 0

0 0.015190
6
0.004642
0.016258
786
0.005265 0.012882 0

0 0.016714
5
0.002592
0.010269
991
0.003987 0.015445 0

0 0.017870
6
0.004386
0.015284
594
0.005796 0.014126 0

0 0.017789
4
0.003128
0.010969
058
0.003304 0.017451 0

0 0.020019
6
0.002129
0.016050
998
0.004577 0.015922 0

0 0.019665
2
0.002074
0.019328
928
0.005449 0.015186 0

0 0.019698
5
0.003209
0.012278
553
0.003616 0.018118 0

0 0.021049
8
0.002556
207

0.013160
6
0.004437 0.018101 0

0 0.021232
2
0.003191
0.018325
435
0.005176 0.019745 0

0 0.024083
7
0.001959
0.006668
374
0.002296 0.02192 0

0 0.023474
5
0.000772
0.026726
828
0.00826 0.021048 0

0 0.027481 0.005817
0.022324
742
0.01354 0.171555 0

0 0.176740
8
0.005277
0.015458
057
0.012396 0.176408 0

0 0.180380
9
0.00278
0.011474
39
0.008216 0.196123 0

0 0.199065 0.004801
0.013016
794
0.00952 0.207981 0

0 0.211039
2
0.003148
0.007262
996
0.007081 0.215109 0

0 0.217207 0.002506
0.036025
366
0.020411 0.229153 0

0 0.236595
4
0.005148
0.014111
285
0.011733 0.245432 0

0 0.248706
9
0.003432
0.026855
83
0.012636 0.251938 0

0 0.258834
8
0.003625
0.028780
479
0.014126 0.263518 0

0 0.270734
1
0.003433
0.023179
273
0.009879 0.275459 0

0 0.281056
1
0.002946
0.037666
081
0.018649 0.287889 0

0 0.295575
1
0.00815
0.017512
725
0.008613 0.299055 0

0 0.303604
4
0.003501
0.017022
215
0.009917 0.314675 0

0 0.319227
1
0.00452
0.010334
556
0.011469 0.323763 0

0 0.326562
6
0.003687
0.009558
087
0.006116 0.339536 0

0 0.342106
2
0.002513
0.020690
446
0.00756 0.017716 0

0 0.022857
6
0.002945
0.011995
957
0.005484 0.023405 0

0 0.026134
9
0.003068
0.013843
296
0.005151 0.024039 0

0 0.027602
6
0.003364
0.010361
34
0.003834 0.032293 0

0 0.034741
6
0.002024
0.010500
249
0.005199 0.034381 0

0 0.036903
1
0.002391
0.011470
447
0.004454 0.038313 0

0 0.041243
8
0.002555
208

0.012327
453
0.004739 0.041115 0

0 0.044423
4
0.002017
0.014764
78
0.005103 0.047376 0

0 0.050891
6
0.002676
0.014735
69
0.004968 0.049919 0

0 0.053389
1
0.002991
0.011692
988
0.004502 0.053919 0

0 0.056564
8
0.002306
0.012956
206
0.005247 0.055506 0

0 0.058590
7
0.002903
0.013184
845
0.004666 0.058335 0

0 0.061442
2
0.002677
0.008734
533
0.006135 0.087656 0

0 0.090843
9
0.0023
0.015593
368
0.007627 0.083005 0

0 0.087033
4
0.002878
0.020724
704
0.007406 0.083473 0

0 0.088213
7
0.002098
0.007432
228
0.027388 0.095744 0

0 0.097604
9
0.002339
0.015655
091
0.033369 0.106573 0

0 0.109600
6
0.003518
0.025698
014
0.027537 0.115145 0

0 0.119510
9
0.007076
0.020457
076
0.080302 0.170517 0

0 0.175072
8
0.005881
0.016399
462
0.041185 0.197135 0

0 0.200528
4
0.002646
0.024729
522
0.090805 0.222217 0

0 0.230608
9
0.012381
0.087396
662
0.067075 0.351877 1 2 0.013802 0.366109
8
0.022718
0.067091
45
0.091788 0.468056 1 4 0.028346 0.479268
9
0.017324
0.034526
036
0.061707 0.627442 0

0 0.638080
9
0.005609
0.022823
726
0.04561 0.758391 0

0 0.764697
5
0.003452
0.014237
288
0.035984 0.782868 0

0 0.792384
5
0.001674
0.039603
149
0.080871 0.745513 0

0 0.752399
6
0.004315
0.237989
275
0.257199 0.695138 1 3.5 0.054794 0.711471
8
0.053557
0.047637
533
0.038456 0.778496 0

0 0.788517
8
0.002182
0.043862
464
0.149563 0.932395 0

0 0.946284
8
0.011447
0.011740
51
0.003619 0.018055 0

0 0.020764
6
0.002428
209

0.013434
328
0.005308 0.019017 0

0 0.022191
1
0.002633
0.012440
273
0.003996 0.019567 0

0 0.022462
1
0.002922
0.013294
985
0.004504 0.02105 0

0 0.024246
6
0.002323
0.020215
788
0.005747 0.023203 0

0 0.027967
8
0.00231
0.016726
65
0.005147 0.024864 0

0 0.028825
8
0.00097
0.012184
609
0.003968 0.026322 0

0 0.029180
4
0.00333
0.016624
373
0.00565 0.029592 0

0 0.033467
2
0.005172
0.019199
855
0.005357 0.030893 0

0 0.035380
2
0.002047
0.013341
068
0.00482 0.046534 0

0 0.049666 0.00397
0.012660
419
0.004989 0.050202 0

0 0.053158
6
0.003163
0.013025
625
0.005429 0.049867 0

0 0.052983
2
0.00378
0.008585
552
0.004643 0.05539 0

0 0.057396
3
0.001949
0.018142
946
0.005842 0.057569 0

0 0.061878
5
0.005437
0.011849
839
0.003407 0.05797 0

0 0.060730
5
0.001276
0.021112
182
0.019734 0.022759 0

0 0.026525
2
0.002916
0.037775
859
0.027731 0.021945 0

0 0.028073
2
0.004144
0.014096
106
0.023367 0.027346 0

0 0.031237 0.001656
0.027346
805
0.029057 0.033528 0

0 0.037865
4
0.004911
0.008876
917
0.022019 0.037227 0

0 0.039734
3
0.002783
0.020772
061
0.030081 0.039197 0

0 0.042858
1
0.003287
0.006492
121
0.024158 0.047541 0

0 0.050542
9
0.002805
0.017125
184
0.033486 0.048668 0

0 0.052215
3
0.004954
0.005797
968
0.021633 0.080617 0

0 0.082367
5
0.00171
0.009384
149
0.029733 0.091031 0

0 0.093201
9
0.004284
0.027620
347
0.025877 0.1732 0

0 0.179739
2
0.002649
210

0.013831
061
0.046391 0.178972 0

0 0.185520
1
0.006378
0.039812
376
0.047201 0.236407 0

0 0.243974
2
0.008615
0.048103
419
0.027349 0.293597 0

0 0.305315
2
0.004703
0.096681
881
0.119966 0.335237 1 4.25 0.036088 0.368727
8
0.011761
0.017364
197
0.004559 0.006131 0

0 0.010179
4
0.001449
0.010329
478
0.003698 0.006836 0

0 0.009241
4
0.003075
0.004285
54
0.002252 0.009915 0

0 0.010924
3
0.002473
0.012562
419
0.004059 0.008255 0

0 0.011221
8
0.003623
0.014755
274
0.004654 0.009089 0

0 0.012558
6
0.002247
0.009145
371
0.002889 0.010777 0

0 0.012935 0.003281
0.012278
338
0.003775 0.009019 0

0 0.011948
8
0.00226
0.017838
227
0.004921 0.010879 0

0 0.015084
3
0.00205
0.013457
635
0.003889 0.010508 0

0 0.013656
8
0.002587
0.013972
412
0.004287 0.011965 0

0 0.015282
8
0.002753
0.014396
124
0.004254 0.013097 0

0 0.016489
5
0.003323
0.017079
851
0.005652 0.012229 0

0 0.016380
3
0.002853
0.008899
1
0.003097 0.014552 0

0 0.016614
3
0.00291
0.016654
752
0.004786 0.013272 0

0 0.017193
5
0.00171
0.010571
271
0.003884 0.0144 0

0 0.016905
7
0.004092
0.015844
504
0.027325 0.039018 0

0 0.043857
1
0.003481
0.026127
844
0.022451 0.048897 0

0 0.053184
9
0.005508
0.062681
175
0.062735 0.20638 0

0 0.222675
1
0.021439
0.137221
24
0.073345 0.228705 0

0 0.250397
9
0.032853
0.160707
298
0.116239 0.247364 0

0 0.270564
7
0.045895
0.060158
605
0.05628 0.26535 1 2.5 0.012082 0.277902
5
0.013025
211

0.022977
525
0.030914 0.321957 1 4.25 0.013887 0.325072
8
0.003345
0.148065
201
0.073807 0.384774 0

0 0.416755 0.014021
0.162498
841
0.115007 0.505896 1 1.75 0.023337 0.555232
8
0.026206
0.022414
953
0.068236 0.547203 1 4 0.03171 0.572723
2
0.006711
0.090340
73
0.037761 0.533208 0

0 0.558026
8
0.003413
0.042415
344
0.035458 0.589931 1 5 0.055614 0.604306
4
0.011244
0.074415
998
0.094829 0.806962 2 1 0.042743 0.824758
8
0.027045
0.332897
812
0.257529 1.070986 0

0 1.147873
5
0.067128
0.095688
401
0.061365 1.120323 0

0 1.154461
6
0.000682
0.030684
684
0.038819 0.253569 0

0 0.260036
4
0.009431
0.009457
647
0.025193 0.267463 0

0 0.271378
8
0.002705
0.059869
682
0.073258 0.664846 1 3.25 0.01431 0.682723
9
0.014877
0.213884
271
0.281105 1.186715 2 2 0.059118 1.217882
1
0.066082
0.200880
129
0.244187 1.780697 1 4.5 0.12207 1.836374
8
0.019487
0.029774
368
0.075515 2.202963 1 3.75 0.029722 2.214152
5
0.022673
0.236443
078
0.18579 2.494195 1 3.5 0.035414 2.556819
9
0.040865
0.378158
976
0.423431 2.552166 1 1 0.057973 2.666762
1
0.061661
0.090922
167
0.086784 2.624609 1 4 0.018525 2.651632
2
0.011213
0.070766
364
0.078066 2.595144 0

0 2.612762
9
0.001889
0.079608
122
0.074676 2.806595 1 4.75 0.01237 2.821908
6
0.004066
0.186843
655
0.267953 2.914298 0

0 3.110050
2
0
0.110502
672
0.114566 2.894677 1 5 0.164194 2.980229 0.00688
0.486807
064
0.455761 3.145252 1 1 0.095373 3.270265
8
0.097401
0.245571
977
0.334313 3.217132 2 1 0.176388 3.286442
8
0.036933
0.011584
966
0.003805 1.003649 0

0 1.006461
1
0.002654
212

0.015723
65
0.004415 1.00157 0

0 1.005277
2
0.002306
0.016509
833
0.004434 1.003062 0

0 1.006949
9
0.002187
0.015061
419
0.004368 1.002365 0

0 1.005917
3
0.001951
0.011630
515
0.003971 1.000883 0

0 1.003598
5
0.003383
0.012165
442
0.003416 1.002066 0

0 1.004961
4
0.001571
0.017000
027
0.005378 1.000951 0

0 1.004968
2
0.003093
0.013580
941
0.004327 1.000779 0

0 1.003970
4
0.002997
0.021763
813
0.00636 1.000412 0

0 1.005491
2
0.003017
0.013111
989
0.004692 1.0016 0

0 1.004750
2
0.004014
0.014032
467
0.004118 1.002109 0

0 1.005403
2
0.002746
0.014333
86
0.004692 1.00206 0

0 1.005382
5
0.003962
0.019471
056
0.006071 1.001032 0

0 1.005690
1
0.003792
0.015707
355
0.004376 1.001621 0

0 1.005347
3
0.002086
0.013990
438
0.00445 1.003514 0

0 1.006796
5
0.003118
0.019366
895
0.006975 0.038196 0

0 0.042742
7
0.005488
0.010954
118
0.003001 0.042796 0

0 0.045302
3
0.001572
0.015855
888
0.004908 0.051404 0

0 0.055192
1
0.003752
0.011428
433
0.003308 0.062377 0

0 0.065091
6
0.003027
0.016907
654
0.005579 0.06368 0

0 0.067612
9
0.004571
0.010549
58
0.004777 0.069712 0

0 0.072215
1
0.002308
0.012833
925
0.004035 0.076356 0

0 0.079339
6
0.002089
0.009504
028
0.002871 0.082282 0

0 0.084554
8
0.001466
0.011953
882
0.003774 0.088793 0

0 0.091574
8
0.00251
0.019994
884
0.006199 0.093752 0

0 0.098531
9
0.003509
0.009148
752
0.002715 0.10012 0

0 0.102381
2
0.002101
213

0.012572
821
0.00369 0.111415 0

0 0.114366 0.00174
0.007489
906
0.003653 0.114929 0

0 0.116811
2
0.001834
0.037911
153
0.028025 0.124071 1 3.25 0.026734 0.132826
1
0.03047
0.017697
821
0.005709 0.153714 0

0 0.157798
8
0.00432
0.047212
321
0.060657 2.101622 1 3.75 0.075714 2.109692
9
0.023582
0 0.016992 2.352693 0

0 2.390916
1
0.001655
0.546644
193
0.386516 2.689205 2 2.5 0.147711 2.818386
6
0.126788
0.466132
513
0.218439 3.624638 1 3 0.074415 3.757323
5
0.075782
0.160378
265
0.166754 4.591564 0

0 4.663338
1
0.03679
0.195197
643
0.16227 6.761556 0

0 6.867943
1
0.097908
1.505619
511
0.919504 10.82229 0

0 11.04872
5
0.651274
0.895241
668
0.66516 9.910224 2 1.5 0.039081 10.23568
6
0.029109
5.384441
339
2.128952 11.7946 1 2.5 0.201011 13.11130
2
0.52025
0.631205
805
1.56754 12.45877 0

0 12.89459
2
0.147426
3.480868
191
1.233868 11.768 1 2.5 0.114969 12.49464
8
0.21835
4.507325
932
1.282981 11.38965 0

0 12.31200
3
0.458408
0.710405
159
0.306736 12.17847 1 4.75 0.085187 12.31870
7
0.040151
0.994162
297
0.38417 11.65228 0

0 11.94383
4
0
0.455286
43
0.35028 11.97618 1 3 0.0418 12.07115
9
0.047459
0.014208
898
0.023596 0.024457 0

0 0.027238
3
0.00399
0.007257
927
0.012851 0.025732 0

0 0.028706
1
0.002315
0.007663
341
0.009124 0.031029 0

0 0.033374
3
0.001485
0.016649
256
0.023506 0.03363 0

0 0.037541
2
0.004578
0.016451
506
0.008647 0.038387 0

0 0.041505
5
0.001688
0.028594
831
0.040666 0.038039 0

0 0.044222
1
0.006506
214

0.010542
358
0.026589 0.04309 0

0 0.046581
5
0.004191
0.013237
252
0.027447 0.056885 0

0 0.061020
6
0.00428
0.016372
631
0.020711 0.089792 0

0 0.094399
6
0.005874
0.015880
778
0.020004 0.167337 0

0 0.173381 0.005203
0.027723
649
0.028378 0.226647 0

0 0.234816
5
0.005295
0.106886
998
0.106849 0.254799 1 2.75 0.034542 0.268005
9
0.037314
0.073299
161
0.326399 0.366594 1 2.5 0.011871 0.383720
2
0.060745
0.036562
245
0.057622 0.449063 1 2.25 0.012875 0.455899 0.013247
0.051338
922
0.077345 0.550061 1 2 0.01211 0.561360
6
0.021137
0.089742
636
0.116866 0.505039 0

0 0.529100
1
0.029612
0.104837
192
0.077118 0.638985 1 1.5 0.031448 0.660588
6
0.032537
0.064347
683
0.080176 0.809014 1 3.75 0.025668 0.821822
9
0.023385
0.238657
366
0.172438 0.962407 1 1.5 0.07529 1.006971
6
0.075757
0.147018
783
0.14969 1.231258 1 1 0.045152 1.263471
8
0.045791
0.219581
749
0.192183 1.360148 2 2.5 0.064706 1.388876
9
0.048519
0.084679
899
0.148899 1.555105 1 4 0.010878 1.586905
4
0.011954
0.306439
801
0.334959 1.605896 1 2.5 0.084166 1.671238
1
0.086211
0.280001
78
0.309464 1.65117 1 3 0.091592 1.671664
7
0.081438
0.189866
569
0.205753 1.780021 2 2.75 0.142337 1.823962
9
0.052079
0.198173
944
0.220128 1.844579 1 1.75 0.029409 1.893589
7
0.048405
0.016261
4
0.232489 1.93212 1 4.75 0.08622 1.994398
9
0.018795
0.177652
133
0.172225 2.058467 2 1.25 0.029288 2.114770
5
0.019578
0.238576
345
0.376824 2.036909 1 3.75 0.105499 2.074855
5
0.087509
0.340874
949
0.281687 2.025006 1 3 0.05489 2.108119
4
0.056448
0.013170
249
0.014993 0.103524 0

0 0.106277
7
0.003977
215

0.009157
592
0.008642 0.112047 0

0 0.113893
6
0.002201
0.022590
157
0.018288 0.163301 0

0 0.168023
2
0.004086
0.005304
765
0.010719 0.183012 0

0 0.185609
4
0.002275
0.013844
186
0.021594 0.195596 0

0 0.200295
6
0.003266
0.025395
381
0.019681 0.209642 0

0 0.214688
1
0.004392
0.007553
13
0.012631 0.220364 0

0 0.224052
4
0.002384
0.019442
436
0.030545 0.231623 0

0 0.236634
1
0.005124
0.012954
416
0.013476 0.246065 0

0 0.250188
6
0.002417
0.029491
978
0.015635 0.254581 0

0 0.260023
5
0.002521
0.006380
758
0.009631 0.27465 0

0 0.277566 0.002736
0.021790
347
0.016658 0.282037 0

0 0.286442
5
0.003316
0.023175
933
0.017906 0.288074 0

0 0.292161
8
0.003945
0.063847
496
0.085226 0.314546 1 4 0.028233 0.329287
4
0.016594
0.405004
963
0.250002 0.450218 1 2 0.098745 0.492853
6
0.099253
0.200704
092
0.440411 3.036 2 3.5 0.050875 3.083383
1
0.055567
0.150441
39
0.119607 3.13747 1 2.75 0.016922 3.192886
5
0.018134
0.770908
344
1.052866 4.793376 3 1.75 0.296972 4.917319
3
0.214
0.881350
926
0.715762 4.730878 2 1 0.114545 4.944457
9
0.112042
1.512828
919
0.739397 4.75647 2 2.25 0.076861 5.122239
9
0.06402
2.946103
704
1.618797 4.473916 3 1 0.218479 5.064863
9
0.178923
2.833385
28
1.241693 4.303079 4 1.25 0.111489 5.012124
3
0.066359
0.735202
043
0.602182 4.562556 2 1.5 0.099977 4.763854
5
0.072601
1.250313
762
1.038954 4.400911 2 1 0.184152 4.680898
7
0.101733
1.544042
353
1.951993 4.264625 3 3 0.200657 4.648142
4
0.138358
0.902230
722
1.062864 4.422655 2 1 0.104653 4.698718
7
0.106548
216

0.923036
947
0.71216 4.594436 1 1.75 0.085969 4.801719
8
0.092174
1.181006
632
0.830198 4.464248 2 3 0.14041 4.679504
4
0.136018
1.100392
803
0.872609 4.344961 3 2 0.110828 4.611249
3
0.108838
1.376934
621
0.99319 8.577326 1 4.5 0.01875 9.195510
5
0.030518
0.008228
279
0.049689 1.093722 0

0 1.096662
4
0.003777
0.015715
021
0.045541 1.099288 0

0 1.102220
9
0.004616
0.018684
937
0.033314 1.109727 0

0 1.114019
9
0.002466
0.008411
285
0.027011 1.133305 0

0 1.139383
4
0.000812
0.015824
258
0.027964 1.170148 0

0 1.174674
7
0.004479
0.150265
001
0.12741 1.196725 1 3.5 0.013911 1.243394
5
0.014509
0.052494
6
0.077015 1.376474 0

0 1.409681
3
0.002524
0.224563
637
0.23953 1.355029 2 1.5 0.132442 1.376896 0.04258
0.448915
139
0.671618 1.678354 1 1 0.089539 1.806607
3
0.092643
0.036902
545
0.154082 1.595319 0

0 1.609610
7
0.015069
0 0.053785 1.70225 0

0 1.712040
9
0.000959
0.657959
526
0.812249 1.779685 0

0 1.983857
4
0.072028
0.393598
523
0.371535 1.708318 1 4 0.085708 1.774595
7
0.051885
0.770934
772
1.44155 1.945198 0

0 2.242624
9
0.145224
0.105460
896
0.19742 1.992771 1 4.5 0.060154 2.027967
9
0.005276

HR hr-sub stress
level
stress-
sub
temp temp-
sub
satisfacti
on
motivati
on
stress concentr
ate
prefere
nce
77 0 30 0 27.761
36
0

74 -3 24 -6 27.775
05
0.0136
89
-1 -1 0 1 dislike
77 0 32 2 27.715
56
-
0.0458
1 -1 0 2 like
217

76 -1

27.658
9
-
0.1024
6
0 -2 -1 -1 dislike
78 1 36 6 27.653
15
-
0.1082
1
1 0 1 1 like
77 0 41 11 27.734
34
-
0.0270
2
-1 -1 -1 -1 dislike
77 0 31 1 27.876
77
0.1154
1
1 -1 2 2 like
76 -1 42 12 27.986
16
0.2248
03
1 0 1 1 like
74 -3 20 -10 28.093
25
0.3318
85
-1 -2 0 -1 dislike

12 -18 28.218
18
0.4568
2
-1 -1 0 -1 dislike
73 -4 22 -8 28.271
74
0.5103
77
-2 -1 -2 -1 dislike
78 1 41 11 28.349
64
0.5882
79
-2 -1 -1 -1 dislike
80 3 38 8 28.413
82
0.6524
59
-1 -1 -1 -2 dislike
79 2 42 12 28.428
95
0.6675
89
1 -1 0 1 like
80 3 42 12 28.419
6
0.6582
44
0 0 -1 1 dislike
54 0 0 0 28.010
9
0

53 -1

28.179
69
0.1687
94
1 0 -1 1 like
51 -3

28.302
28
0.2913
79
2 0 1 1 like

28.241
41
0.2305
1
2 1 2 -1 like
60 6 3 3 28.120
69
0.1097
89
1 0 0 1 like
72 18 11 11 28.102
98
0.0920
84
2 1 2 1 like

24 24 28.240
48
0.2295
75
2 1 2 0 like
74 20 18 18 28.421
3
0.4104 2 1 2 1 like

14 14 28.638
46
0.6275
59
1 0 2 0 like
69 15 3 3 28.848
45
0.8375
5
0 -1 1 -2 dislike
66 12 2 2 28.895
07
0.8841
66
1 1 1 1 like
59 5 1 1 28.934
63
0.9237
25
1 0 0 0 like
218

59 5 1 1 29.017
49
1.0065
92
1 1 2 -1 like
60 6

29.173
41
1.1625
14
1 1 1 1 like
59 5

29.285
46
1.2745
59
-2 -2 -2 -2 dislike
85 0 70 0 29.806
15
0

90 5 73 3 29.813
03
0.0068
83
1 2 1 1 like
87 2 69 -1 29.779
84
-
0.0263
1
0 1 1 0 like
86 1 66 -4 29.767
97
-
0.0381
8
1 1 0 1 like
90 5 66 -4 29.776
6
-
0.0295
5
0 1 1 0 like
91 6 51 -19 29.825
41
0.0192
6
1 1 1 0 like
90 5 64 -6 29.884
75
0.0786
04
1 2 2 1 like
89 4 68 -2 29.962
11
0.1559
65
1 2 1 1 like
90 5 59 -11 30.017
34
0.2111
94
1 0 1 0 like

63 -7 30.055
16
0.2490
14
-1 -1 -1 -1 dislike
88 3 57 -13 30.089
34
0.2831
94
1 1 0 0 like
93 8 75 5 30.176
72
0.3705
71
2 2 2 1 like
92 7 77 7 30.225
25
0.4190
96
2 2 1 1 like
89 4 65 -5 30.277
28
0.4711
29
2 2 1 1 like
88 3 72 2 30.343
02
0.5368
66
-1 -1 -1 -1 dislike
68 0 5 0 30.284
92
0

67 -1 13 8 30.233
11
-
0.0518
-1 -1 -1 2 dislike
72 4 11 6 30.155
92
-0.129 1 1 1 2 like

14 9 30.010
69
-
0.2742
2
1 1 2 2 like

13 8 29.773
98
-
0.5109
3
0 1 0 1 like
219

15 10 29.723
95
-
0.5609
7
1 2 1 2 like
73 5 14 9 30.008
54
-
0.2763
7
2 2 1 2 like
70 2 18 13 30.231
38
-
0.0535
3
1 1 1 1 like
70 2 16 11 30.483
9
0.1989
82
1 0 0 0 like

11 6 30.761
82
0.4769
05
1 1 -1 -2 dislike
71 3 16 11 31.032
91
0.7479
94
0 0 -1 0 dislike
69 1

31.312
87
1.0279
56
1 1 2 1 like
71 3

31.540
93
1.2560
11
-1 -1 -1 -1 dislike

17 12 31.697
35
1.4124
38
1 1 1 1 like
72 4 17 12 31.805
97
1.5210
53
-2 -2 -2 -1 dislike
108 0 92 0 29.248
9
0

95 -13 84 -8 29.443
92
0.1950
18
1 0 -1 0 dislike
99 -9 89 -3 29.697
21
0.4483
13
1 1

1 like
104 -4 90 -2 29.832
67
0.5837
72
1 1 0 1 like
101 -7 88 -4 30.022
07
0.7731
66
-1 -1 -2 -1 dislike
99 -9 89 -3 30.203
87
0.9549
69
0 0 0 0 dislike
97 -11 90 -2 30.330
85
1.0819
52
2 2 1 1 like

88 -4 30.418
72
1.1698
17
0 0 1 0 like
100 -8 85 -7 30.490
28
1.2413
79
0 -1 0 0 dislike
105 -3 91 -1 30.627
64
1.3787
39
-1 -1 -2 -2 dislike
103 -5 91 -1 30.724
51
1.4756
08
1 1 1 1 like
104 -4 91 -1 30.777
83
1.5289
29
-1 -1 -1 -1 dislike
105 -3 91 -1 30.799
33
1.5504
28
-1 -1 -1 -1 dislike
97 -11 84 -8 30.827
64
1.5787
36
1 0 1 0 like
220

94 -14 85 -7 30.880
08
1.6311
82
0 0 0 0 dislike
70 0 11 0 32.570
2
0

69 -1

32.601
02
0.0308
16
1 2 1 1 like
70 0

32.677
41
0.1072
1
1 1 0 1 like
69 -1

32.902
21
0.3320
13
1 2 1 2 like
69 -1

32.988
31
0.4181
11
-1 -1 -1 -1 dislike
72 2

33.029
61
0.4594
07
2 2 2 2 like
71 1

33.077
25
0.5070
46
1 -1 0 1 like
69 -1

33.160
43
0.5902
26
1 1 0 1 like
67 -3

33.261
93
0.6917
34
1 1 0 -1 like
71 1

33.269
18
0.6989
8
2 2 2 2 like
73 3

33.280
69
0.7104
89
-1 -1 -1 -1 dislike
72 2

33.271
31
0.7011
11
1 0 0 0 like

33.425
23
0.8550
3
2 2 1 2 like
69 -1

33.519
54
0.9493
41
-1 -1 1 -1 dislike
69 -1

33.617
15
1.0469
48
1 1 1 1 like
64 0 4 0 30.757 0

62 -2

30.714
47
-
0.0425
3
2 2 1 1 like

30.559
49
-
0.1975
1
2 2 2 1 like
69 5

30.484
91
-
0.2720
9
2 2 2 2 like
61 -3

30.417
49
-
0.3395
1
2 2 2 2 like
60 -4

30.382
56
-
0.3744
4
1 1 1 0 like
221

59 -5

30.410
32
-
0.3466
8
2 2 2 2 like

30.445
56
-
0.3114
4
1 1 2 2 like

30.463
13
-
0.2938
7
1 1 0 1 like
60 -4

30.505
4
-
0.2516
-1 -1 -1 -2 dislike
63 -1

30.558
39
-
0.1986
1
0 1 0 1 like
64 0

30.575
49
-
0.1815
1
-1 -1 -1 -1 dislike
63 -1

30.530
86
-
0.2261
4
-1 -1 0 0 dislike
71 7

30.509
84
-
0.2471
6
1 1 2 2 like

30.520
25
-
0.2367
5
1 2 1 2 like
65 0

27.657
43
0

63 -2

27.722
71
0.0652
72
2 2 1 1 like

27.702
12
0.0446
9
2 2 1 2 like
68 3

27.693
92
0.0364
9
2 1 1 2 like

27.681
64
0.0242
03
2 2 1 2 like

27.650
04
-
0.0074
2 2 2 2 like

27.641
75
-
0.0156
8
2 2 2 2 like
71 6

27.634
56
-
0.0228
7
2 2 2 2 like
72 7

27.645
58
-
0.0118
5
1 1 1 1 like
68 3

27.625
35
-
0.0320
9
-1 -2 -2 -1 dislike
222

67 2

27.546
22
-
0.1112
1
2 1 1 1 like
68 3

27.458
73
-
0.1987
1
-1 -2 -2 -1 dislike
63 -2

27.405
09
-
0.2523
5
2 1 1 1 like

27.338
39
-
0.3190
4
2 1 2 1 like
66 1

27.249
87
-
0.4075
6
0 -1 -1 -1 dislike
72 0 0 0 30.754
4
0

72 0 11 11 31.029
19
0.2747
88
1 1 1 1 like
70 -2 5 5 31.362
55
0.6081
53
2 2 2 2 like
70 -2 4 4 31.622
21
0.8678
13
2 2 2 2 like
73 1 12 12 31.842
8
1.0884 -1 -1 -1 0 dislike
75 3 3 3 32.056
28
1.3018
84
0 0 0 0 dislike
74 2 9 9 32.259
57
1.5051
71
2 2 2 2 like
77 5 6 6 32.416
37
1.6619
67
1 1 1 1 like

15 15 32.583
85
1.8294
5
0 0 1 1 like
78 6 13 13 32.775
87
2.0214
68
-1 -1 -1 -1 dislike
73 1 6 6 32.927
06
2.1726
58
1 1 1 1 like
76 4 8 8 33.043
52
2.2891
16
-1 0 -1 0 dislike

6 6 33.127
97
2.3735
71
1 1 1 1 like
80 8 15 15 33.216
82
2.4624
2
0 0 1 0 like

79 0 18 0 31.511
65
0

80 1 21 3 31.721
24
0.2095
9
1 0 -1 1 like
76 -3 12 -6 31.909
78
0.3981
32
2 2 2 2 like
223

73 -6 14 -4 32.106
51
0.5948
56
1 1 2 1 like
73 -6 9 -9 32.302
28
0.7906
3
1 0 0 0 like
75 -4 9 -9 32.535
2
1.0235
54
2 1 2 1 like
76 -3 16 -2 32.699
32
1.1876
66
2 2 2 1 like

16 -2 32.850
26
1.3386
1
2 2 1 2 like
73 -6 8 -10 33.002
36
1.4907
09
1 1 1 1 like

11 -7 33.074
29
1.5626
44
0 0 -2 0 dislike
73 -6 8 -10 33.114
74
1.6030
88
2 2 2 2 like
76 -3 17 -1 33.120
11
1.6084
64
0 0 -1 0 dislike
80 1 16 -2 33.159
18
1.6475
33
1 0 -1 1 like
75 -4 13 -5 33.227
93
1.7162
79
2 2 2 2 like
75 -4 15 -3 33.277
95
1.7662
96
-1 -1 -2 -1 dislike
110 0 87 0 31.402
57
0

110 0 90 3 31.744
45
0.3418
83
0 -1 0 -1 dislike
105 -5 83 -4 32.139
16
0.7365
94
1 2 -1 1 like
107 -3 84 -3 32.484
83
1.0822
67
2 2 -1 2 like
106 -4 85 -2 32.739
79
1.3372
21
-1 -2 0 -1 dislike

83 -4 32.916
05
1.5134
84
0 -1 1 1 like
106 -4 84 -3 33.060
49
1.6579
27
2 2 0 2 like
105 -5 84 -3 33.202
14
1.7995
69
2 2 1 2 like
106 -4 86 -1 33.334
47
1.9319
01
-1 0 -1 1 dislike
111 1 87 0 33.432
59
2.0300
2
-1 -2 -1 -2 dislike
106 -4 82 -5 33.508
52
2.1059
49
1 1 -1 1 like

85 -2 33.572
61
2.1700
41
1 1 0 1 like
107 -3 84 -3 33.635
57
2.2330
07
0 -1 1 -1 dislike
224

87 0 33.713
75
2.3111
83
1 2 1 2 like
106 -4 84 -3 33.780
66
2.3780
89
0 0 1 0 like
86 0 30 0 27.150
33
0

83 -3

27.119
24
-
0.0311
0 -1 -1 2 dislike
85 -1

27.101
23
-
0.0491
1
1 0 1 1 like
79 -7

27.225
8
0.0754
7
-1 -2 -1 0 dislike

27.544
46
0.3941
3
2 2 0 2 like
83 -3 30 0 27.788
12
0.6377
83
2 2 2 2 like
88 2

27.959
72
0.8093
83
1 1 1 2 like
80 -6

28.079
76
0.9294
29
0 0 1 -1 dislike
85 -1

28.173
35
1.0230
17
1 2 1 1 like
86 0 43 13 28.251
18
1.1008
44
0 -1 -1 1 dislike
87 1 30 0 28.330
48
1.1801
45
0 -2 2 -2 dislike

36 6 28.397
79
1.2474
53
1 1 2 2 like
92 6 38 8 28.496 1.3456
67
1 -1 2 0 like
89 3 39 9 28.661
55
1.5112
18
-1 -2 -2 -2 dislike
89 3 42 12 28.910
35
1.7600
22
-1 -1 -1 -1 dislike
81 0 10 0 24.965
9
0

82 1 9 -1 25.400
21
0.4343
1
-1 -1 0 -2 dislike
86 5 51 41 25.911
36
0.9454
56
0 0 -1 -1 dislike
87 6 42 32 26.470
08
1.5041
79
0 0 -1 -1 dislike
94 13 55 45 27.158
79
2.1928
89
1 0 0 0 like
89 8 53 43 27.551
03
2.5851
3
1 1 1 0 like
88 7 32 22 28.028
32
3.0624
19
1 0 1 0 like
83 2 32 22 28.509
25
3.5433
5
0 1 0 -1 dislike
225

89 8 57 47 28.887
77
3.9218
66
1 1 0 -1 like
92 11 64 54 29.243
4
4.2774
97
-1 -1 -1 -1 dislike
91 10 59 49 29.633
72
4.6678
21
-1 -1 -1 -1 dislike
93 12 65 55 29.888
1
4.9221
97
0 0 0 0 dislike
91 10 62 52 30.498
43
5.5325
27
0 -1 -1 -1 dislike
96 15 62 52 31.135
93
6.1700
26
0 -1 0 -1 dislike
95 14 64 54 31.289
43
6.3235
26
0 -1 0 -1 dislike
69 0 3 0 28.099 0

79 10 11 8 28.163
91
0.0649
08
2 2 2 1 like

12 9 28.232
98
0.1339
82
1 2 2 2 like

8 5 28.211
17
0.1121
67
1 1 2 1 like
81 12 12 9 28.211
92
0.1129
22
-1 -1 -1 -1 dislike
94 25 55 52 28.167
11
0.0681
14
1 1 0 0 like
96 27 53 50 28.101
25
0.0022
5
2 2 2 1 like
95 26 70 67 28.050
48
-
0.0485
2
1 1 2 1 like
95 26 63 60 28.022
98
-
0.0760
2
2 2 2 2 like

62 59 27.980
5
-
0.1185
-1 -1 -1 -1 dislike
93 24 63 60 27.962
4
-
0.1366
-1 -1 -1 -2 dislike

63 60 27.925
2
-
0.1738
1 1 2 1 like
92 23 60 57 27.935
57
-
0.1634
3
1 1 2 1 like

54 51 27.928
22
-
0.1707
8
2 2 2 2 like
98 29 62 59 27.942
2
-
0.1568
2 2 2 2 like
60 0 0 0 28.142
2
0

226

60 0

28.321
45
0.1792
52
1 1 0 1 like
61 1

28.548
52
0.4063
22
0 1 0 1 like
62 2

28.766
03
0.6238
35
0 0 -1 1 dislike
63 3

28.966
23
0.8240
33
1 1 0 1 like
68 8

29.149
29
1.0070
88
1 1 0 1 like
63 3

29.327
71
1.1855
1
1 2 1 -1 like
67 7

29.443
15
1.3009
45
1 1 1 1 like
60 0

29.521
29
1.3790
88
0 1 0 1 like

29.573
03
1.4308
34
-1 0 -1 0 dislike
63 3

29.666
82
1.5246
15
1 1 1 1 like
61 1

29.711
71
1.5695
12
-1 0 -1 0 dislike
64 4

29.773
24
1.6310
42
0 -1 -1 -1 dislike
62 2

29.810
65
1.6684
5
0 1 1 1 like
60 0

29.860
77
1.7185
7
-1 -1 0 1 dislike
92 0 50 0 27.914
45
0

93 1 61 11 28.392
51
0.4780
56
1 1 1 2 like
93 1 57 7 29.304
49
1.3900
38
-1 -2 -1 -2 dislike

64 14 30.095
77
2.1813
22
0 1 1 1 like

54 4 30.440
47
2.5260
17
2 1 1 2 like

66 16 30.710
58
2.7961
26
2 2 2 2 like
93 1 74 24 31.156
75
3.2423
02
0 -1 -1 1 dislike

74 24 31.529
94
3.6154
88
-1 -1 -1 -1 dislike
98 6 73 23 31.791
33
3.8768
83
1 1 0 1 like

73 23 31.919
92
4.0054
72
2 1 1 2 like

77 27 31.969
63
4.0551
75
-1 -2 -1 -2 dislike
227

71 21 32.020
51
4.1060
56
1 1 0 1 like
101 9 75 25 32.060
84
4.1463
88
2 1 1 1 like
100 8 75 25 32.116
38
4.2019
29
-1 -1 -1 -2 dislike
99 7 75 25 32.171
52
4.2570
75
2 2 0 1 like
88 0 56 0 30.482
08
0

94 6 63 7 30.610
95
0.1288
63
1 0 0 1 like
93 5 36 -20 30.738
57
0.2564
84
2 2 2 1 like
87 -1 37 -19 30.847
3
0.3652
15
-1 -2 0 -1 dislike
86 -2 50 -6 30.967
89
0.4858
06
1 2 -1 1 like
87 -1 45 -11 31.096
99
0.6149
05
0 0 0 -1 dislike

30 -26 31.214
38
0.7322
96
0 1 -1 0 dislike
87 -1 40 -16 31.304
87
0.8227
83
1 2 1 2 like
86 -2 26 -30 31.411
65
0.9295
69
2 2 2 2 like
85 -3 28 -28 31.469
79
0.9877
05
1 1 -1 0 like
85 -3 28 -28 31.526
99
1.0449
03
-2 -2 -2 -2 dislike
90 2 36 -20 31.617
89
1.1358
08
0 0 0 0 dislike
90 2 35 -21 31.703
06
1.2209
81
2 2 -1 1 like
93 5 36 -20 31.761
47
1.2793
91
0 0 0 0 dislike
90 2 25 -31 31.812
92
1.3308
35
-1 -1 0 0 dislike
71 0 7 0 27.535
35
0

73 2 4 -3 27.518
56
-
0.0167
9
1 2 2 1 like

2 -5 27.491
36
-
0.0439
9
0 -2 1 -1 dislike

3 -4 27.453
27
-
0.0820
8
1 0 0 1 like
228

72 1 3 -4 27.343
33
-
0.1920
2
-1 -1 -1 -1 dislike
69 -2 4 -3 27.160
71
-
0.3746
4
2 1 2 1 like
68 -3 2 -5 27.050
7
-
0.4846
5
1 0 1 1 like
67 -4 2 -5 27.038
93
-
0.4964
2
0 -1 -1 -1 dislike
70 -1 1 -6 27.208
13
-
0.3272
2
1 1 0 0 like
74 3 5 -2 28.521
38
0.9860
27
1 1 0 1 like
73 2 3 -4 28.528
02
0.9926
66
-2 -2 -2 -2 dislike
72 1 1 -6 28.515
18
0.9798
3
1 0 0 -1 dislike
71 0 1 -6 28.484
2
0.9488
47
1 0 0 -1 dislike

10 3 28.501
46
0.9661
09
1 0 0 -1 dislike

2 -5 28.575
82
1.0404
7
1 0 1 1 like
86 0 30 0 29.871
23
0

101 15 68 38 30.029
93
0.1587
01
2 1 -1 0 like
84 -2 36 6 30.237
93
0.3667
01
1 1 1 1 like
80 -6 19 -11 30.477
26
0.6060
29
1 0 0 -1 dislike

7 -23 30.653
98
0.7827
5
1 1 0 1 like
75 -11 7 -23 30.749
54
0.8783
08
1 -1 -1 1 dislike

1 -29 30.88 1.0087
67
2 1 0 1 like
74 -12 5 -25 31.019
34
1.1481
11
0 -1 -1 -1 dislike
87 1 21 -9 31.130
66
1.2594
22
1 1 1 1 like
89 3 49 19 31.198
54
1.3273
08
0 -1 -2 -1 dislike
85 -1 50 20 31.423
91
1.5526
73
1 0 0 0 like
85 -1 50 20 31.516
39
1.6451
6
0 -1 1 1 like
229

87 1 48 18 31.757
85
1.8866
19
-1 -1 -1 -1 dislike
84 -2 33 3 31.866
21
1.9949
8
1 -1 1 1 like
89 3 53 23 31.897
03
2.0257
99
1 1 0 1 like
109 0 89 0 30.634
4
0

108 -1 88 -1 30.427
13
-
0.2072
7
-1 -1 0 0 dislike

87 -2 30.469
03
-
0.1653
7
1 1 2 1 like
108 -1 84 -5 30.641
93
0.0075
34
1 1 2 1 like
103 -6 75 -14 30.792
13
0.1577
31
0 -1 0 0 dislike
104 -5 82 -7 30.964
98
0.3305
84
0 0 0 0 dislike

88 -1 31.160
18
0.5257
8
2 1 2 1 like
109 0 90 1 31.524
92
0.8905
18
2 2 1 1 like

87 -2 31.778
58
1.1441
83
-1 -1 -1 -1 dislike

88 -1 31.979
61
1.3452
07
-2 -2 -2 -2 dislike
109 0 89 0 32.220
9
1.5865
02
-1 -1 -1 -1 dislike

89 0 32.377
25
1.7428
46
-1 -1 0 0 dislike
107 -2 86 -3 32.470
21
1.8358
08
-1 -1 0 0 dislike
99 -10 81 -8 32.551
54
1.9171
44
1 1 0 1 like
102 -7 86 -3 32.647
82
2.0134
2
-1 -1 -1 0 dislike
84 0 31 0 28.807
05
0

81 -3 35 4 28.945
97
0.1389
17
0 1 0 1 like
86 2 50 19 29.246
21
0.4391
6
2 1 1 -1 like
92 8 40 9 29.427
05
0.6200
01
1 0 -1 0 dislike
90 6 35 4 29.547
49
0.7404
42
-1 -1 0 -2 dislike
92 8 51 20 29.663
35
0.8563
05
0 1 0 1 like
230

45 14 29.838
18
1.0311
33
1 -2 2 -1 dislike

42 11 30.102
23
1.2951
76
1 1 2 1 like
88 4 49 18 30.290
65
1.4836 0 -1 0 1 dislike
93 9 51 20 30.406
4
1.5993
53
-1 -1 2 -1 dislike
91 7 62 31 30.439
95
1.6329
01
1 0 1 1 like
90 6 59 28 30.427
59
1.6205
4
0 -1 1 -2 dislike

48 17 30.430
15
1.6230
98
-1 -1 0 0 dislike
89 5 66 35 30.444
21
1.6371
63
0 1 0 -1 dislike
91 7 56 25 30.434
41
1.6273
6
-2 -2 1 -1 dislike
93 0 67 0 33.060
03
0

95 2 68 1 33.116
69
0.0566
55
2 2 2 1 like
92 -1 59 -8 33.060
31
0.0002
78
1 -2 -1 2 dislike
93 0 57 -10 32.916
28
-
0.1437
5
1 1 0 1 like
97 4 70 3 32.844
25
-
0.2157
9
0 -1 -1 2 dislike
97 4 60 -7 32.750
48
-
0.3095
6
2 2 2 1 like
94 1 66 -1 32.751
33
-
0.3087
1
1 1 0 1 like
95 2 69 2 32.774
34
-
0.2856
9
1 1 0 1 like
94 1 66 -1 32.702
54
-
0.3574
9
2 2 2 2 like
96 3 66 -1 32.593
77
-
0.4662
6
1 2 -2 2 like
99 6 65 -2 32.583
97
-
0.4760
7
0 -2 2 0 dislike
90 -3 51 -16 32.575
44
-
0.4845
9
2 0 2 1 like
231

90 -3 57 -10 32.574
16
-
0.4858
7
2 2 2 2 like

39 -28 32.562
16
-
0.4978
8
-2 -2 -2 -1 dislike
85 -8 50 -17 32.555
84
-
0.5042
1 1 0 1 like
82 0 31 0 28.917
3
0

79 -3 23 -8 28.974 0.0567 0 1 1 1 like
80 -2 15 -16 29.245
02
0.3277
16
-1 -1 -1 -1 dislike
83 1 17 -14 29.568
77
0.6514
74
-2 -1 -1 -1 dislike
85 3 35 4 29.986
7
1.0694
05
1 1 0 1 like
86 4 25 -6 30.175
62
1.2583
23
2 1 -2 2 like
85 3 33 2 30.492
51
1.5752
08
0 0 0 0 dislike

33 2 30.630
26
1.7129
62
0 0 0 0 dislike
75 -7 14 -17 30.949
74
2.0324
38
1 1 1 1 like
82 0 22 -9 31.098
91
2.1816
11
1 1 1 1 like
85 3 40 9 31.201
1
2.2837
98
-2 -2 -2 -2 dislike
88 6 45 14 31.281
74
2.3644
38
-1 -1 -1 -2 dislike
82 0 27 -4 31.366
45
2.4491
5
1 0 0 -1 dislike
84 2 24 -7 31.405
45
2.4881
55
0 0 -1 -1 dislike
85 3 46 15 31.408
02
2.4907
16
1 1 0 0 like
82 0 12 0 30.450
4
0

87 5 21 9 30.704
03
0.2536
33
1 1 0 1 like

14 2 31.238
64
0.7882
39
0 0 -1 0 dislike
83 1 13 1 31.564
7
1.1143
05
2 1 1 2 like
77 -5 6 -6 31.718
77
1.2683
7
1 1 1 1 like

5 -7 31.889
97
1.4395
67
2 2 1 2 like
232

78 -4 21 9 32.131
46
1.6810
59
0 0 -1 0 dislike

20 8 32.393
44
1.9430
43
0 0 -1 0 dislike

10 -2 32.710
79
2.2603
87
1 1 1 1 like
82 0 6 -6 32.976
26
2.5258
62
0 -1 1 -1 dislike

22 10 33.150
33
2.6999
28
0 1 0 0 like

28 16 33.475
34
3.0249
44
0 -1 0 -1 dislike
84 2 34 22 33.597
59
3.1471
9
0 0 -1 0 dislike
82 0 23 11 33.725
44
3.2750
43
0 0 0 0 dislike
85 3 26 14 33.839
38
3.3889
77
2 1 1 0 like
74 0 15 0 30.403
28
0

73 -1 0 -15 30.590
87
0.1875
86
1 1 1 -1 like
77 3 12 -3 31.111
66
0.7083
72
2 -2 0 -2 dislike
77 3 15 0 31.298
41
0.8951
27
2 2 2 2 like
83 9 17 2 31.420
2
1.0169
13
-1 -1 0 0 dislike
78 4 21 6 31.601
67
1.1983
89
2 2 1 2 like
77 3 23 8 31.776
7
1.3734
22
2 2 2 1 like
78 4 15 0 31.904
08
1.5007
99
1 1 -1 1 like
81 7 10 -5 31.956
7
1.5534
22
2 2 2 1 like
74 0 9 -6 31.972
9
1.5696
18
1 2 -1 2 like
76 2 14 -1 31.973
75
1.5704
71
-2 -2 -2 -2 dislike
74 0 20 5 31.948
18
1.5448
97
-1 -1 -1 0 dislike
75 1 11 -4 31.925
87
1.5225
86
-1 -1 1 0 dislike
76 2 12 -3 31.926
69
1.5234
05
1 2 2 2 like
72 -2 13 -2 31.914
38
1.5110
94
-2 0 2 -1 dislike
96 0 72 0 28.851
6
0

233

99 3 72 0 29.092
08
0.2404
82
2 2 1 2 like
100 4 77 5 29.369
05
0.5174
49
2 2 2 1 like
99 3 72 0 29.653
38
0.8017
77
1 2 2 2 like
99 3 70 -2 29.871
79
1.0201
87
0 1 0 1 like
100 4 76 4 30.038
18
1.1865
8
1 1 1 1 like
99 3 70 -2 30.191
26
1.3396
62
2 1 2 1 like
92 -4 69 -3 30.366
14
1.5145
36
0 1 1 0 like
98 2 74 2 30.445
53
1.5939
34
1 2 0 1 like
104 8 77 5 30.557
15
1.7055
48
0 0 1 0 like
100 4 74 2 30.738
64
1.8870
39
-1 -1 0 -1 dislike

75 3 30.869
97
2.0183
67
-1 0 0 0 dislike
98 2 69 -3 30.970
54
2.1189
41
2 1 -1 2 like
102 6 73 1 31.036 2.1844 2 2 -1 1 like
99 3 72 0 31.136
43
2.2848
26
2 2 -1 1 like
85 0 48 0 32.884
23
0

83 -2 41 -7 33.009
74
0.1255
04
2 1 1 0 like
82 -3 61 13 33.287
23
0.4029
96
2 2 2 2 like
109 24 85 37 33.531
48
0.6472
42
1 0 1 0 like

88 40 33.762
13
0.8778
98
2 1 2 2 like

89 41 33.942
39
1.0581
54
1 1 1 1 like
103 18 84 36 34.060
92
1.1766
86
2 2 2 2 like

81 33 34.081
95
1.1977
17
0 0 0 1 like

81 33 34.087
07
1.2028
32
1 0 0 0 like
102 17 75 27 34.060
02
1.1757
83
-1 -2 -1 -2 dislike

78 30 34.035
97
1.1517
34
0 0 0 0 dislike
234

78 30 34.039
26
1.1550
29
2 1 1 -1 like
100 15 78 30 34.083
77
1.1995
4
1 1 1 1 like
99 14 76 28 34.116
21
1.2319
8
1 1 1 1 like
92 7 75 27 34.098 1.2137
67
-2 -2 -2 0 dislike
103 0 60 0 32.097
57
0

99 -4 58 -2 32.128
48
0.0309
09
1 0 0 1 like
102 -1 64 4 32.209
03
0.1114
66
0 -1 1 0 dislike
101 -2 66 6 32.314
21
0.2166
46
1 1 -1 1 like

74 14 32.435
21
0.3376
46
0 0 1 0 like
98 -5 71 11 32.631
59
0.5340
23
1 0 0 -1 dislike
99 -4 66 6 32.805
89
0.7083
19
2 1 0 1 like
104 1 77 17 33.023
7
0.9261
38
1 1 0 0 like
105 2 72 12 33.228
2
1.1306
3
0 1 -1 1 like
111 8 78 18 33.445
97
1.3484
01
1 1 -1 1 like
110 7 80 20 33.657
23
1.5596
63
-1 -1 0 -2 dislike
107 4 76 16 33.809
02
1.7114
5
0 0 0 0 dislike
105 2 75 15 33.961
57
1.8639
99
0 0 0 0 dislike
104 1 75 15 34.116
8
2.0192
36
-1 0 0 -1 dislike
106 3 77 17 34.242
54
2.1449
74
1 1 -1 1 like
114 0 87 0 32.882
5
0

117 3 84 -3 32.950
38
0.0678
77
1 1 0 1 like
113 -1 85 -2 33.123
89
0.2413
85
-1 -1 -1 -1 dislike
114 0 85 -2 33.208
82
0.3263
2
0 0 -1 0 dislike
117 3 87 0 33.329
39
0.4468
93
1 1 1 1 like
112 -2 87 0 33.422
66
0.5401
56
2 2 2 1 like
235

110 -4 87 0 33.442
79
0.5602
87
0 -1 0 0 dislike
115 1 89 2 33.449
18
0.5666
8
1 1 1 1 like
115 1 93 6 33.508
28
0.6257
79
0 0 -1 1 dislike
116 2 87 0 33.527
03
0.6445
33
-1 -1 -1 -2 dislike
115 1 91 4 33.551
75
0.6692
54
-2 -2 -2 -2 dislike
116 2 88 1 33.564
82
0.6823
21
1 1 1 1 like
114 0 87 0 33.568
38
0.6858
77
-1 0 -1 -1 dislike
113 -1 84 -3 33.564
11
0.6816
07
-1 -1 -1 -1 dislike
112 -2 81 -6 33.557
72
0.6752
21
1 1 1 1 like
76 0 10 0 31.168
73
0

75 -1 6 -4 31.488
34
0.3196
11
-1 0 0 1 dislike
73 -3 3 -7 32.027
39
0.8586
6
1 1 1 2 like
72 -4 13 3 32.470
16
1.3014
31
1 2 2 1 like
77 1 5 -5 32.740
89
1.5721
52
0 0 0 1 like
82 6 10 0 32.958
46
1.7897
26
1 1 0 1 like
80 4 -2 -12 33.252
93
2.0841
94
2 2 2 2 like
94 18 15 5 33.354
36
2.1856
27
1 1 1 1 like
78 2 4 -6 33.425
34
2.2566
11
1 0 1 0 like
82 6 19 9 33.556
3
2.3875
62
-2 -2 -2 -2 dislike
77 1 5 -5 33.651
77
2.4830
37
2 2 2 2 like
83 7 4 -6 33.736 2.5672
67
0 0 -1 0 dislike
85 9 5 -5 33.796
95
2.6282
17
-1 -1 0 0 dislike
75 -1 3 -7 33.872
02
2.7032
83
1 0 1 1 like
76 0 10 0 33.900
7
2.7319
72
-2 -2 -2 -1 dislike

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

IEQ, sleep quality, and IoT: meta-analysis on improving IEQ and sleep quality using IoT

PDF

Human-building integration based on biometric signal analysis: investigation of the relationships between human comfort and IEQ in a multi-occupancy condition

PDF

Data visualization in VR/AR: static data analysis in buildings

PDF

Exploration for the prediction of thermal comfort & sensation with application of building HVAC automation

PDF

Indoor air quality for human health in residential buildings

PDF

Using bio-signals with smart windows

PDF

Enhancing thermal comfort: data-driven approach to control air temperature based on facial skin temperature

PDF

Visualizing architectural lighting: creating and reviewing workflows based on virtual reality platforms

PDF

Development of data-driven user-centered building façade design guideline models: machine learning-based approaches to predict user preferences

PDF

Human–building integration: machine learning–based and occupant eye pupil size–driven lighting control as an applicable visual comfort tool in the office environment

PDF

Developing environmental controls using a data-driven approach for enhancing environmental comfort and energy performance

PDF

Human-building integration: Investigation of human eye pupil sizes as a measure of visual sensation in the workstation environment

PDF

BIM+AR in architecture: a building maintenance application for a smart phone

PDF

Building bridges: filling gaps between BIM and other tools in mechanical design

PDF

Office floor plans generation based on Generative Adversarial Network

PDF

Development of AI-driven architectural design guidelines: establishing human biometric signal-driven architectural design guideline as a function of psychological principles

PDF

Economizer performance and verification: effect of human behavior on economizer efficacy and thermal comfort in southern California

PDF

Daylighting study of a LEED platinum laboratory building: a post-occupancy evaluation comparing performance in use to design intent

PDF

Net zero energy building: the integration of design strategies and PVs for zero-energy consumption

PDF

Daylighting in Riyadh and Los Angeles: comparison of cultural factors in potential market penetration

Asset Metadata

Creator Chen, Dejian (author)

Core Title Quantify human experience: integrating virtual reality, biometric sensors, and machine learning

School School of Architecture

Degree Master of Building Science

Degree Program Building Science

Degree Conferral Date 2022-05

Publication Date 04/15/2024

Defense Date 03/09/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag architectural design,biometric data,biosensor,indoor environment,machine learning,OAI-PMH Harvest,virtual reality

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Choi, Joon-Ho (committee chair), Kensek, Karen (committee member), Liu, Yan (committee member)

Creator Email chendejian1993@gmail.com,dejianch@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC110964824

Unique identifier UC110964824

Document Type Thesis

Format application/pdf (imt)

Rights Chen, Dejian

Type texts

Source 20220416-usctheses-batch-926 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu