University of Southern California Dissertations and Theses

Utilizing 311 service requests as a signature of urban location in the City of Los Angeles

(USC Thesis Other)

Utilizing 311 service requests as a signature of urban location in the City of Los Angeles

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Utilizing 311 Service Requests as a Signature of
Urban Location in the City of Los Angeles

by

Richard Windisch

A Thesis Presented to the
Faculty of the USC Graduate School
University of Southern California
In Partial Fulfillment of the
Requirements for the Degree
Master of Science
(Geographic Information Science and Technology)

August 2019

ii

Copyright © 2019 by Richard Windisch

iii

I dedicate this paper to my parents, my sister, and to all of my family and friends who have been
with me throughout this entire process. Without your guidance and support, none of this would
have been possible.
iv
Table of Contents
List of Figures ............................................................................................................................... vii
List of Tables ................................................................................................................................. ix
Acknowledgements ......................................................................................................................... x
List of Abbreviations ..................................................................................................................... xi
Abstract ......................................................................................................................................... xii
Chapter 1 Introduction .................................................................................................................... 1
1.1. Motivation ...........................................................................................................................2
1.2. Research Question ..............................................................................................................4
1.3. Study Area ..........................................................................................................................5
1.3.1. Los Angeles Demographics .......................................................................................7
1.4. Origins of 311 Systems .....................................................................................................14
1.4.1. New York .................................................................................................................15
1.4.2. Philadelphia ..............................................................................................................16
1.5. MyLA311 and Selected Service Requests ........................................................................18
1.5.1. MyLA311 .................................................................................................................18
1.5.2. Illegal Dumping .......................................................................................................21
1.5.3. Graffiti Removal Requests .......................................................................................23
1.5.4. Homeless Encampments ..........................................................................................24
1.5.5. Dead Animal Removal .............................................................................................27
1.5.6. Broken Streetlights ...................................................................................................28
1.6. MyLA311 as the Start of Los Angeles Open Data ...........................................................29
1.7. Thesis Structure ................................................................................................................30
Chapter 2 Literature Review ......................................................................................................... 31
2.1. Volunteered Geographic Information ...............................................................................31
v
2.1.1. Ease of Acquisition and Collection of Data .............................................................31
2.1.2. Spatial Data Quality .................................................................................................33
2.2. Open Data .........................................................................................................................36
2.2.1. Shifting Towards Open Data ...................................................................................36
2.2.2. Public Participatory GIS and E-governance ............................................................37
2.3. Analyzing 311 Service Requests ......................................................................................41
2.4. Multivariate Clustering .....................................................................................................43
2.5. Collective Efficacy ............................................................................................................46
Chapter 3 Data and Methodology ................................................................................................. 50
3.1. Methods Overview ............................................................................................................50
3.2. Data and Processing ..........................................................................................................51
3.2.1. MyLA311 .................................................................................................................52
3.2.2. American Community Survey Data .........................................................................54
3.2.3. Contextual Boundary Data .......................................................................................58
3.3. Data Aggregation ..............................................................................................................58
3.3.1. Scale of Analysis ......................................................................................................58
3.3.2. Data Normalization ..................................................................................................59
3.4. Data Analysis ....................................................................................................................60
3.4.1. Processing and Joining Data to Shapefiles ..............................................................61
3.4.2. Multivariate Clustering Analysis .............................................................................62
3.4.3. Sociodemographic Analysis .....................................................................................63
Chapter 4 Results .......................................................................................................................... 65
4.1. Analysis of Service Requests ............................................................................................65
4.2. Service Request Clusters from MyLA311 Data ...............................................................68
4.2.1. 311 Clusters and Resulting Neighborhoods .............................................................72
vi
4.3. Sociodemographic Attributes of Resulting 311 Clusters ..................................................73
4.3.1. Sociodemographic Cluster Characteristics ..............................................................74
Chapter 5 Conclusions and Discussion ......................................................................................... 78
5.1. Significance of Findings ...................................................................................................78
5.1.1. Proof of Concept ......................................................................................................79
5.1.2. Implications for Los Angeles ...................................................................................79
5.2. Study Limitations and Future Research ............................................................................80
5.2.1. Limitations ...............................................................................................................80
5.2.2. Future Research .......................................................................................................81
5.3. Conclusion ........................................................................................................................82
References ..................................................................................................................................... 84

vii
List of Figures
Figure 1 Neighborhoods comprising Los Angeles ......................................................................... 6
Figure 2 Distribution of race and ethnicity ..................................................................................... 8
Figure 3 Distribution of highest educational attainment ............................................................... 10
Figure 4 Maps of mean income .................................................................................................... 11
Figure 5 Distribution of uninsured and people living below poverty ........................................... 13
Figure 6 Unemployment rate ........................................................................................................ 14
Figure 7 Philly 311 ....................................................................................................................... 17
Figure 8 City of Los Angeles MyLA311 Service website ............................................................ 19
Figure 9 MyLA311 mobile phone application interface ............................................................... 20
Figure 10 LA Sanitation promoting reporting illegal dumping .................................................... 22
Figure 11 Graffiti on a building in Los Angeles ........................................................................... 23
Figure 12 Homeless encampment on public right of way ............................................................ 25
Figure 13 Stages of E-government ............................................................................................... 39
Figure 14 Social media post encouraging self-reporting by constituents ..................................... 41
Figure 15 K-Means Cluster Function ........................................................................................... 45
Figure 16 Methodology process .................................................................................................... 51
Figure 17 MyLA311 2018 point data ........................................................................................... 53
Figure 18 Map detailing the amount of 311 requests for service ................................................. 66
Figure 19 Chart detailing the sources of MyLA 2018 service request data .................................. 68
Figure 20 Multivariate Clustering of 311 Data ............................................................................. 69
Figure 21 Average Service Requests per 311 Cluster .................................................................. 70
Figure 22 Radar chart of 311 service request cluster characteristics ............................................ 71
viii
Figure 23 Multivariate Clustering Box Plots – 311 Data ............................................................. 72
Figure 24 Radar Chart of Sociodemographic data from 311 Clustering ...................................... 75

ix
List of Tables
Table 1 All Possible Service Request Types from MyLA311 ...................................................... 21
Table 2 Data sets used in the analysis ........................................................................................... 52
Table 3 MyLA311 Data Attributes ............................................................................................... 54
Table 4 American Community Survey Census Datasets .............................................................. 56
Table 5 Standardizing in Excel ..................................................................................................... 60
Table 6 Software required for analysis ......................................................................................... 61
Table 7 Multivariate Clustering Table Results – 311 Data ........................................................... 72
Table 8 Z-Scores of Cluster Characteristics ................................................................................. 76

x
Acknowledgements
I am grateful to my committee chair, Dr. Lisa Sedano, and committee members, Dr. An-Min Wu
and Dr. Su Jin Lee, for their patience, guidance, and feedback during the writing and analysis of
this manuscript. I am grateful to Beau MacDonald for providing feedback on my ideas and
supporting me through my endeavors during my undergraduate and graduate research. I am
grateful to Dr. Noli Brazil for introducing me to the 311 data and its potential research
applications. My employer, LA Sanitation, and Cecile Buncio and Bryan Cowitz were very
generous in allowing me to be flexible with my work schedule, earning my gratitude and
appreciation. Finally, I want to extend my gratitude to my parents, family, and friends for
supporting my educational pursuits and always being there for me. Without the help of all of
you, this feat would not have been possible.

xi
List of Abbreviations
ACS American Community Survey
BSS Building of Street Services
FOSS Free and Open Source Software
GIS Geographic information system
LAHSA Los Angeles Homeless Services Authority
LAPD Los Angeles Police Department
LASAN Los Angeles Sanitation
MyLA311 My Los Angeles 311
PPGIS Public Participatory GIS
SR Service Requests
TIGER Topologically Integrated Geographic Encoding and Referencing
VGI Volunteered Geographic Information

xii
Abstract
In order to increase citizen engagement, in 2013, the City of Los Angeles introduced the
MyLA311 application, a smartphone app that allows residents to easily request city services.
Previous service requests were funneled through four separate data service management systems
and lacked transparency; the improved centralized system increases public data access and
efficiency, all while ensuring a uniform tracking methodology across all departments. Citizens
act as agents creating data each time a service request is made. Ease of reporting and increased
use of mobile applications or digital platforms to track and monitor service requests creates huge
volumes of volunteered geographic information (VGI) data. Los Angeles’s shift towards open
data supports data-driven decision-making regarding city services and mitigation of problems.
While the original purpose behind the push towards publicly accessible information was
accountability, a new purpose for the data was found in the possibility of constituents creating
additional insights. The attributes of VGI provided through the MyLA311 service requests were
analyzed to determine fitness for use in spatial analysis. Los Angeles experiences a great deal of
spatial heterogeneity given the differences in socioeconomic attributes and local neighborhood
contexts. Distinct signatures of the local urban context, similar to a neighborhood, are
determined through a multivariate cluster analysis of MyLA311 Service Requests and
sociodemographic data at the census tract level. This spatial analysis provides stakeholders and
civic leaders with insights into which physical problems need focus in certain geographically
defined communities detailed in the results and conclusion.

1

Chapter 1 Introduction
This thesis spatially analyzes 311 requests for service in the City of Los Angeles and uses them
to identify patterns of reporting across the city. The resulting data consists of daily requests for
service and are a form of volunteered geographic information (VGI), as it is spatial data directly
sourced from citizens. The service request data is publicly available to promote accountability,
data-driven governance and civic improvement. With freely accessible data, the opportunities for
analysis are plentiful. At the time of writing, a search in Google Scholar for “MyLA311” returns
not even two full pages of results. Of all of these results, only one regards a spatial analysis of
the data; all remaining results mainly speak of MyLA311 as a step in the right direction for cities
and open data. The analysis of 311 data herein is unique to Los Angeles and informed by
previous 311 data analyses from other cities that detail the limitations and demonstrate different
use cases. The purpose of this thesis is to identify spatial signatures, or a unique set of
characteristics about a place, across Los Angeles through multivariate cluster analyses of 311
service requests and local sociodemographic features from census data.
With the City of Los Angeles now sharing 311 data of service requests via its online open
data platform, an opportunity has arisen to perform data analysis on requests for service
submitted by citizens. While many different types of service requests are available for analysis,
this thesis focuses on service requests for illegal dumping, graffiti removal, homeless
encampments, dead animal removal, and broken streetlights. This decision stems from the
frequency of selected service request types and through the requests’ shared public/visual
characteristics of visibility in an urban environment. The analysis utilizes census tracts in order
to provide a stable set of geographic units for the presentation of statistical data. While Los
Angeles is divided into Council Districts and neighborhoods within local contexts, a multivariate
2

cluster analysis of service requests on the scale of the census tract provides an informative basis
for determining clusters of similarity. As Waldo Tobler’s first law of geography states,
“everything is related to everything else, but near things are more related than distant things”
(Tobler 1970). Analysis of publicly available MyLA311 data provides a use case for the
introduction of viable tools for informed understanding of urban spaces and city management
practices.
1.1. Motivation
The collection and management of massive data quantities about communities is no
longer viewed as an impediment to the implementation of GIS services and analyses, but an
accessible process through technological advances and mobile mapping technology (Novak
1995). VGI offers previously unimaginable sources and quantities of data. By reporting their
own issues, citizens have the chance to play a part in managing their built environment while
enacting the ideas of community betterment. With citizens as sensors, cities can explore a
powerful and comprehensive approach reaching beyond the digital divide of citizens and
government (Goodchild 2011). Through engaging citizens in the process of acquiring and
utilizing geographic information, VGI has the potential to alter this landscape significantly and
soften criticisms of citizen engagement in local government as inaccessible or difficult to
accomplish (Toregas 2001).
This project builds on the recent shift in municipal government to be more data-driven
and to prioritize publicly available data. The current Mayor of Los Angeles, Eric Garcetti, and
the City Controller, Ron Galperin, emphasize the importance of setting benchmarks for
accountability of city services, as well as constant improvement (Los Angeles City Controller
2019). The data produced by the city is then built into open access online dashboards to monitor
3

the progress, providing a snapshot into how the city is functioning. Both constituents and city
officials can track the progress of goals online through the Mayor’s Dashboard. The Los Angeles
GeoHub and Open Data Portal, two publicly available online civic infrastructures, cater to city
employees and residents in offering data and resources. The GeoHub is the city’s online platform
for exploring, visualizing, and downloading geographic information system (GIS) based data.
City leaders promote “civic hacking,” a term in which “hacking” refers to perverting government
data’s original purpose to solve a different civic problem (Tauberer 2014). The idea driving civic
hacking is public engagement with the data to solve problems that matter to them. Analysis of
civic data can result in direct, fact-based influences in policy with more economic, sustainable
and community-based outcomes.
The MyLA311 website provides spatiotemporal data that can be used for a multitude of
analyses using a GIS. Through an internal data management system, MyLA311 cost-effectively
manages and collects data from citizens self-reporting through online forms, phone calls, or
requests within an application on mobile devices. In addition to providing useful metrics for
civic accountability, analysis of VGI from service requests provides information not previously
known about an area when taken into account with other data sets. The location component of a
service request is the most integral aspect of the data because it ties the spatiotemporal aspects of
a service request to a unique location, qualifying it for use in a GIS analysis. Within this thesis,
the temporal and spatial components of the MyLA311 data support spatial analysis while also
providing a descriptive comparison of the geospatial information of the 311 data to the
sociocultural context of Los Angeles neighborhoods.
4

1.2. Research Question
A spatial signature is a unique set of characteristics about a place, as defined in Wang et
al. (2017). The 311 data analyzed is comprised of service requests for illegal dumping, graffiti
removal, homeless encampments, dead animal removal and broken streetlights. The
sociodemographic features from the census data focus on population diversity, education and
income and employment. Spatially distributed patterns emerge through multivariate clustering of
data across a defined area. An understanding of the nuances of working with VGI and MyLA311
data informs the thesis to determine differences in signatures, which may reflect the geographic
constraints of similar characteristics that make up neighborhoods in Los Angeles.
After completion of this analysis, comparisons can be drawn between the distribution of
cluster locations and the arbitrarily defined political boundaries existing in Los Angeles. For
example, stakeholders can incorporate their own location with the resulting cluster outcomes to
see how politically defined geographic political regions compare with the actual urban makeup
of an area. Additionally, civic entities can observe characteristics about the people of a place and
the services requested for problems in an area. While the research does not answer why different
signatures exist, it supports policy and decision makers in how they can better understand the
context of an area, allowing them to better determine an approach for service.
This thesis therefore answers a two-part research question:
1. What clusters of neighborhoods can be identified by spatial clusters of common
patterns of 311 service requests?
2. Do the areas identified as clusters share similar sociodemographic features?
5

1.3. Study Area
The study area for the analysis is the City of Los Angeles. MyLA311 only contains
service requests within the Los Angeles city boundaries. This analysis would work at the county
level if neighboring cities used a similar methodology for dissemination of service request data
and had parallel request types. The land area of the city is 468.67 square miles and contains a
population of roughly four million people (U.S. Census Bureau 2018). While Los Angeles is one
large city, it is often said to be a collection of many smaller neighborhoods, each with a unique
identity. These neighborhoods, as determined by crowdsourced opinions by the Los Angeles
Times, are revealed in Figure 1. The names of some of these neighborhoods are used throughout
this thesis to provide geographical context for the analysis.
6

Figure 1 Neighborhoods comprising Los Angeles, source: Los Angeles Times
7

1.3.1. Los Angeles Demographics
In order to better understand the different characteristics that inform clustering in Los
Angeles, it is important to have a sufficient understanding of the general makeup of the city.
Common sociodemographic features representing important phenomena in population diversity,
education, and income and employment within Los Angeles are examined within the analysis.
Census data validates the analysis and further informs MyLA311 service request clusters to
determine a proxy for the socioeconomic characteristics of neighborhoods. This analysis uses
data from the U.S. Census 2017 5-year estimate of the American Community Survey (ACS) for
socioeconomic and demographic information.
The distribution of race and ethnicity within Los Angeles census tracts by concentration
is demonstrated in Figure 2. The analysis uses the racial and ethnic categories of the 2017 ACS;
more information discussing the reasoning is discussed in Chapter 3. The four maps in Figure 2
highlight the areas with the highest percentage of a single race and/or ethnicity. The Valley, the
Westside, Malibu and Hollywood Hills have the highest percentage of Non-Hispanic White
residents. African American neighborhoods are mostly located within central Los Angeles. Small
pockets of Asian population exist near the Downtown and Midtown areas of Los Angeles.
Hispanics or Latinos comprise the majority ethnic group in Los Angeles and have dense
communities in the northern portion of the Valley, the Downtown area, East Los Angeles, and
the corridor down to San Pedro Harbor.
8

Figure 2 Distribution of race and ethnicity by census tract in Los Angeles
9

Percentages of the highest educational attainment of each census tract are displayed
through the maps in Figure 3. Graduated symbology on a scale from the minimum percentage of
education attainment to the maximum percentage is used to account for the disparities between
each level. Attainment of only a high school level of education is most prevalent in the northern
portion of the Valley, and then from Downtown to the San Pedro Harbor area. College education
at the levels of associates and bachelor’s degrees, are more equally distributed throughout the
city, while the Westside of Los Angeles and Malibu have a higher proportion of residents who
have completed some type of graduate school.
10

Figure 3 Distribution of highest educational attainment in Los Angeles census tracts
11

Two different symbolizations were used to display the Mean Income (Figure 4). The
same data is represented in two different ways due to the disparity in income levels present with
unclassed data symbolization detailing percentages from zero up to $430,000. Through unclassed
symbolization, the wealth appears to be only concentrated in the Hollywood Hills. However,
once the symbolization is by standard deviation, characteristics of overall income in the city
become more apparent. While the Westside and the Hollywood Hills remain with the highest
average incomes, the perimeters of the Valley, East Los Angeles, and pockets in the Harbor area
and Downtown are now visible. The Downtown pocket of higher mean income parallels the
recent shift towards higher incomes in that area as high-end residential development gentrifies
the area (Collins 2016). Otherwise, the general Downtown and Harbor Corridor are in the lower
end of the average income bracket.

Figure 4 Maps of both the unclassed (left) and standard deviation (right) of mean income in Los
Angeles census tracts
12

The percentages of those who are uninsured and those living below the poverty line are
determined through different metrics, yet there is an overlap present between the two groups.
The percentages of residents who are uninsured and who are below the poverty line mirror each
other in some regards (Figure 5). This may be attributed to insurance not being seen as a high
priority when money must go towards other immediate necessities when one is living below the
poverty line (Bundorf 2006). The Census Bureau establishes placement below the poverty line
by a set of income thresholds varying by family size and composition. If a respondent’s stated
total income is less than the family's determined threshold, then that family unit and each
individual in it is considered living below the poverty line (U.S. Census Bureau, 2018).
Economists established trends demonstrating that health insurance is generally affordable
for between one quarter and three quarters of adults who are not insured (Los Angeles County
Department of Public Health 2017). More recently and specific to Los Angeles, LA County
Public Health (2017) determined a significant reduction in the number of uninsured individuals
from 2011 to 2015. The general geographic locations for higher populations both percentage
below the poverty line and percentage uninsured are in the northern portion of the Valley and the
corridor from Downtown Los Angeles to the San Pedro Harbor.
13

Figure 5 Distribution of uninsured and people living below poverty in Los Angeles census tracts
The final census variable observed is the unemployment rate, displayed in Figure 6. The
unemployment rate is the most widely known labor market indicator and reflects the number of
unemployed people as a percentage of the labor force. The official definition of unemployment
encapsulates people who are jobless, actively seeking work and available to take a job. This
statistic is formulated by number of unemployed as a percentage of the labor force (the sum of
the employed and unemployed). Within the map, the darkest spots appear to be outliers: Cal
State Northridge in the Valley, UCLA on the Westside and USC in South Central all have high
rates of unemployment based upon the high residency of students in the area. The darkest, large
census tract north of Downtown is Griffith Park, and the darkest census tract east of Downtown
is Skid Row, known for a high population of homeless individuals.
14

Figure 6 Unemployment rate in Los Angeles by census tract
1.4. Origins of 311 Systems
Prior to the existence of 311 services, citizens only had 911 as an easy to remember
option to call if they needed help, or they could call city departments directly with service
requests if they had the correct phone numbers available. As a result, 911 received an overload of
non-emergency calls, which led to delays in the response time for emergency services attending
to actual emergencies (Schwester 2009). An alternative number for non-emergencies was created
to alleviate the congestion of calls to 911, and 311 was chosen as the number. A successful trial
run of a 311 system in Baltimore resulted in the Federal Communications Commission reserving
311 as a national help number for non-emergency situations, as deemed in Federal
Communications Commission Report 97-51 (1997). The introduction of 311 systems in cities
15

provided an easily accessible, single point of entry to local government information and services
and a “real” person to speak to. Customer service-oriented call centers and the creation of service
data gained momentum to become a catalyst for modernizing city operations. This section
describes the initial implementations of 311 systems in New York and Philadelphia, both cities
Los Angeles sought to emulate in its own implementation.
1.4.1. New York
In 2002, New York City’s Mayor Michael Bloomberg sought out the creation of a
customer service focused initiative to provide access to GIS integrated information about city
complaints from 15 major categories, including air and water quality, construction, noise,
animals, snow, streets and sidewalks, and transit and parking (Nam 2012; Nadeu 2011). New
York, with a population double that of Los Angeles, needed to implement a robust 311 system
with the goals of allowing government departments to focus on respective core missions and
efficiently manage workloads.
Now a modernized and transparent system, NYC 311 provides accurate and consistent
data tracking and analysis of all service requests, with links to other civic resources for
residential engagement. New York’s 311 system is representative of one of the most significant
correlations of citizens engaging with local government, as more than 8 million service requests
are created annually (Kontokosta 2017). Data-driven mindsets changed internal views of the 311
system as a method for complaints into an opportunity for larger-scale solutions; it could now
become an integration of all civic culture, influencing organizational structure, technological
components and workers (Nam 2012). In 2010, Wired Magazine included a feature analyzing
New York’s 311 data and detailing the resulting policy changes. For example, once
geographically identified patterns of service requests for handling excessive noise became the
16

top service request, the administration instituted noise-abatement programs and quelled the
problem (Johnson 2010). As Los Angeles readies itself for scaling up its 311 system and digital
technologies, as noted in Mayor Garcetti’s Directive No. 3 (2013), and it can look to New York
as an example for city service request management and resulting policies.
1.4.2. Philadelphia
According to an analysis by the Pew Charitable Trust using U.S. Census American
Community Survey data, Philadelphia was determined to be the poorest large city in the United
States (Trinacria 2018). In 2008, Philadelphia Mayor Michael Nutter unveiled Philly 311 as a
modernized government-to-constituent customer service model for accountability and to improve
life for the city’s residents. Creating a database of feedback under a single entity was the goal;
previously, the city had multiple customer service hotlines, however each of them was located in
singular compartmental silos. Rather than forcing city departments to merge with the system, the
city made it optional in order to ease the adoption process. Philly 311 had a clear process for
maintenance and corrections in order to maintain its knowledgebase, further bolstering the need
for interdepartmental collaboration, transparency, and integration (Nam 2012). Philly 311
(Figure 7) differentiates itself from the MyLA311 system dashboards by providing a live map
feed of the reported problems.
17

Figure 7 Philly 311 website interface for reporting 311 service requests
Philly 311 enabled data-driven management in the city and its data became essential in
managing, tracking, and monitoring organizational performance. The Philadelphia City council
reported resources became used more effectively, saving money and time of legislative members
to direct towards other important needs of constituents (Nam 2014). While the system was an
improvement, city personnel still struggled to modify the system and reported difficulty in data
management practices. Multiple internal studies of Philly 311 lead to changes and updates in
how the service would be run and continue to adapt to the city’s changing needs. In 2015, Philly
311 modernized to a cloud-based system, effectively eliminating antiquated technology,
upgrading communications, and becoming more “scalable, resilient, transparent and responsive
to the needs of constituents” (Bengfort 2019). Philadelphia’s 311 system provides a model for
Los Angeles to modernize the MyLA311 system.
18

1.5. MyLA311 and Selected Service Requests
Los Angeles’ 311 Call Center provides general information relating to city run programs
and services. Citizens have the option to submit city service requests as needed and are
connected directly with specific city departments, bridging communication between constituents
and the city and simplifying operations with the City of Los Angeles. The system provides an
extensive list of options available for submission as service requests through MyLA311. This
section provides context of case studies on use of service requests within GIS analyses and the
reasoning behind reporting relating to illegal dumping, graffiti removal, homeless encampments,
dead animal removal, and broken streetlights. The reasoning behind reporting specific problems
lends insight to the urban signature, characteristics and values of a location.
1.5.1. MyLA311
Individual Los Angeles city departments had veteran service request systems, yet these
compartmentalized, department-specific silos of information lacked transparency and hindered
achievement of data-driven government. The four custom built existing systems were designed
prior to the development of APIs and system interoperability as required features of online
services. The different systems were unable to exchange information and resulted in duplicates
of service request IDs and miscommunication. Introduced in 2013, the newly comprehensive
MyLA311 system was produced by 3Di Systems. Adaptations to new technological trends,
namely call centers, mobile applications and open data, necessitated updated technological
infrastructure and removal of older system architecture, which obstructed innovation. The new
system interface (Figure 8) utilizes the Oracle WebCenter Portal for enterprise portal, Oracle
WebCenter Content for Customer Management System, and Oracle Siebel for Customer
19

Relationship Management to provide an enterprise portal and user interface capable of serving
constituents and their customer service requests through an integrated system.

Figure 8 City of Los Angeles MyLA311 Service website
In order to meet the needs of individuals from a variety of backgrounds, many methods of
submitting a MyLA311 service request exist. The current options include through a phone call,
city attorney, city council office, self-report by driver, e-mail, fax, letter, request through the
Mayor’s Office, queue initiated customer call, radio request, TDD/Nex Talk (service for deaf and
blind) request, Twitter complaint, voicemail, walk-in to city department, online web form, the
MyLA311 online portal and MyLA311 Mobile Application submission (Figure 9). The
significance of the source of service requests can be identified through analysis of the MyLA311
data to determine which sources are used most frequently for submission of specific service
request types.
20

Figure 9 MyLA311 mobile phone application interface
In order to access the data produced from MyLA311, it is necessary to go through a
separate website hosted by the city called Los Angeles Open Data. This Socrata powered website
hosts a variety of data, with much of it produced by standard city processes and services.
MyLA311 data is available going back to 2013. The MyLA311 publicly available data is not the
same as internal city MyLA311 data. Publicly available datasets from MyLA311 only include
service requests of bulky items, dead animal removal, electronic waste, feedback, graffiti
removal, homeless encampments, illegal dumping pickup, metal/household appliances, multiple
streetlight issue, other, report water waste, and single streetlight issue. The full list of MyLA311
data available through access to the city’s backend is in Table 1; within the table, the five service
request types analyzed in this thesis are underlined. These service request types were selected
21

because the reports are expected to come from people who did not make or cause the problem
being reported.
Table 1 All Possible Service Request Types from MyLA311

Another difference in the publicly available versus internal MyLA311 data is the level of
detail and commentary. The nature of MyLA311’s standardized open data does not include the
unique comments or photos relating to each service request; as a result, additional circumstances
of each request cannot be assumed, and the service request must be accepted as a baseline
service request called in by a citizen. This internal cleaning of the data before publishing it
serves to standardize the information and remove personally identifiable information.
1.5.2. Illegal Dumping
Service requests for illegal dumping seek to rectify the illegal disposing of appliances,
barrels, construction debris, electronics, furniture, household trash, leaking liquids, tires, or yard
debris. Within Los Angeles city boundaries, citizens have the option to call Los Angeles
311 Theme Service Request Type
Animal Related Services Request animal related services including dead animal removal, loose or confined animals.
Investigations
Homeless encampment, Illegal auto repair, Illegal construction, Illegal construction fence, Illegal
discharge of water, Illegal dumping in progress, Illegal excavation, Illegal sign removal, Leaf
blower violation, News rack violation, Non-Compliant Vending, Obstructions and Tables , chairs
obstructing and Report Water Waste
Parks
Park facility and field maintenance, trash and cleanliness issues, security and park
tree/animal/bug issues
Problems & Repairs Report graffiti or issues with streetlights
Refuse & Pickups
Bulky item, Containers, Electronic waste, Metal / household appliances, Illegal dumping, Service
not complete and etc
Sanitation Billing
Bulky Item fee, Extra capacity charge, Solid resource fee and Sewer Service Charge
Adjustments. For recycLA billing issues, go to recycLA.com or call the Customer Care Center at
1-800-773-2489.
Street Problem/Repair
Barricade removal, Bus pad/landing, Curb repair, Flooding, General street inspection,
Guard/warning rail maintenance, Gutter repair, Land/mudslide, Pothole, Sidewalk repair and
Street sweeping
Transportation Dockless Mobility Enforcement
Trees/Vegetation
Bees or beehive, Median island maintenance, Overgrown vegetation/plants, Palm fronds down,
Street tree inspection, Street tree violations, Tree emergency, Tree obstruction, Tree permits and
Weed abatement for pvt parcels
Other
To be used ONLY if the issue being reported does not fit into any of the SR Types available on
this list. Select Radio buttons above to see more SR Types.
Feedback Used for commentary
22

Sanitation and coordinate the pickup of the above items through a free service for either
electronic or bulky item pickup. Service request types for bulky item or electronic pick-ups were
created as a service to individuals removing their own belongings, whereas the illegal dumping
service request type is used for disposal of items in the public realm without clear or legal
ownership. The city and council offices often implore residents to assist in cleaning through
reporting illegal dumping through outreach, such as Figure 10, over social media or pamphlets.

Figure 10 LA Sanitation image on Facebook promoting reporting illegal dumping
If citizens are unable to dispose of their waste in a manner considered easy and low-cost,
illegal dumping is often the outcome, as “dumping is sensitive to the cost of legal waste
management and the threat of enforcement” (Matsumoto 2011). A contributing factor to illegal
dumping may be the shortage of proper, or “accessible,” waste treatment facilities. People are
driven to dispose of their waste through illegal means to avoid the cost of paying for these
services (Ichinose 2011). Consistent monitoring for illegal dumping would be inefficient and
costly in terms of resources, lending weight to the inclusion of illegal dumping as something
citizens can report through community participation. A cleaner environment leads to an
increased quality of life for inhabitants.
23

1.5.3. Graffiti Removal Requests
Service requests for graffiti removal seek to rectify the illegal writing or drawings made
on a wall or other surface, often without permission and within public view. Graffiti occurs on a
variety of surfaces and at a variety of scales, as exemplified by Figure 11. The standardization of
public MyLA311 service requests removes any detail or characteristics of the graffiti. The
MyLA311 graffiti removal request data only entails the graffiti that people care about enough to
report. The visual characteristics of graffiti impact how individuals feel about public safety,
economic development and businesses, the quality of life for individuals in an area and feelings
about neighborhoods.

Figure 11 Graffiti on a building in Los Angeles
Some cultural components of the context surrounding why graffiti occurs extend beyond
borders. Megler (2014) comprehensively analyzes the spatial relationship of graffiti in San
Francisco using a combination of census data and 311 data. She identified geographic and
sociodemographic factors with significant correlation to graffiti reports. Most graffiti occurs
along streets that are arteries because there is the most visual impact possible of graffiti being
24

seen. In regard to a trend with sociodemographic characteristics, more reports will be generated
from a higher income community even if there is the same amount of graffiti in a lower income
community, all other variables equal. This relates to the tolerance for graffiti in an area as part of
an urban signature, yet tolerance cannot be specifically tested for through the graffiti removal
service request data.
1.5.4. Homeless Encampments
A homeless encampment entails one or more shelters consisting from lean-tos made of
cardboard, to tents, to more elaborate structures that are used by homeless individuals. An
encampment has no safe way to store and clean food, attracts disease-carrying vermin through
accumulating trash and often contains biohazards resulting from human waste or makeshift fuel
sources. These poor hygiene conditions are ripe for spreading a multitude of diseases, further
endangering the already marginalized population. Encampments bring about problems in the
homeless population, the environment and the larger community through unhealthy and unsafe
the conditions.
Encampments do not have a set temporal definition or location, despite environmental
factors and the built environment potentially influencing prevalence and longevity of an
encampment. Citizens have the option to report a homeless encampment represented by an
individual repeatedly occupying the same location or establishing a shelter, potentially creating a
public hazard or nuisance such as in Figure 12. After a homeless encampment cleanup service
request is filed, a report is sent to the Bureau of Street Services Investigation and Enforcement
Division, LASAN, Los Angeles Housing Services Authority, and Los Angeles Police
Department. Initial contact must be made by LAHSA offering services to the individuals in an
encampment. Notices of the dates of encampment clean-ups are posted in order to provide a
25

warning of the occurrence and give individuals a chance to relocate. Due to strict adherence to
policy, procedure and scheduling, the process between reporting an encampment and the clean-
up itself can take upwards of 75 days. The city seeks to enforce measures intended to protect a
citizen’s right to safe, healthy and accessible public spaces in an equitable and responsible
manner.

Figure 12 Homeless encampment on public right of way in Los Angeles
Homelessness is a publicly visible effect of poverty and vicious inequality. As more
individuals experience homelessness, their presence becomes more integrated into the physical
structure of a city. Homelessness is associated with a variety of sociodemographic and spatial
factors. When mapped over a geographic location, certain characteristics of homeless individuals
may tend to stand out. For example, in terms of household characteristics, the “presence of
children, age of the head and the head’s drug and alcohol problems” were found to be
significantly associated with the probability of being homeless (Early 2005). Once out on the
street, the location that a homeless individual will frequent may shift towards a location where
their needs can be met, either through outreach, shelters, or policy more favorable to their
26

presence. This leads to the presence of encampments near food, alcohol, employment (or crime)
opportunities, and shelter from the elements (Early 2005).
Generally, reports of encampments stem from opportunity to increase the sense of
personal safety in an area. The community feels the impact of increased transient presence as
impeding on daily life. Locations adjacent to transient encampments experience “higher levels of
petty and serious crime unrelated to [self-described] ‘routine behaviors,’ such as drug dealing
and usage, disturbance, theft, prowling, burglary, panhandling, fighting, vandalism, armed
robbery, rape, and aggravated assault” (Chamard 2010). Public perception of transient activity
may incite feelings of lawlessness in an area, dissuading people from spending time or money in
an area with heavy homelessness as an illegitimate use of public space. Chamard (2010) finds
residents living in close quarters to homeless encampments to “suffer disproportionately from
crime committed by transients.” McCormack (2010) understands that within the characteristics
of urban spaces, “most personal safety concerns mentioned in studies were associated with the
presence of undesirable users,” effectively labeling the homeless as “others.” In this regard,
people may be more likely to report homeless encampments in areas they frequent outside the
realm of their residence or job location in order to push them out of their immediate location due
to concerns about increasing safety.
While an individual’s reasoning behind reporting a homeless encampment is not detailed
in the MyLA311 data, individualist perspective models may offer some insight as to why:
These pathology-based models explain homelessness as a consequence upon deviant
behavior and values arising from an individual’s mental illness, chronic substance abuse,
family disorders, disaffiliation from community, or willful participation in a “culture of
poverty” (Koenig 2007).
Within this perspective, individuals reporting homeless encampments are attempting to prevent
the values they associate with homelessness as values characteristic of the location they reside in.
27

While extreme variations of perspectives on homelessness exist, a general air of NIMBY-ism,
where people would like the problem of homelessness to be solved but at a distance from where
they reside, is attributed to their requests for removal of the encampments (Dear 1992).
1.5.5. Dead Animal Removal
The inclusion of dead animal removal as service requests relates to a desire for a clean
city and lessening potential health risks associated with the carcasses of dead animals. Urban
spaces are constructed in a way to prioritize human mobility, not mobility of animals; as a result,
spatial and constitutive characteristics of a place are aligned in a practical manner for human
purposes through infrastructure (Lulka 2013). Physical contact with a dead animal carcass poses
the chance of transferring diseases detrimental or deadly for humans and domesticated pets,
making timely removal of the dead animal through submission of service requests extremely
important to both the environment and inhabitants.
Data on dead animal removal is “highly indicative of a general pattern of urban relation
between humans and nonhumans” (Lulka 2013) in an urban physical environment. Automobiles
are the medium for the most often occurring interaction of human-caused death of animals.
Mobilization in a car without the direct physical, human bodily movement results in disconnect
between humans and the outside occurrences of a vehicle; it is in this space where road kill
occurs (Lulka 2013). Should a civilian hit an animal, it is unlikely that a driver will take the time
to submit a service request for removal of the dead animal, as the main priority in the moment is
traveling to the destination.
Lulka (2013) finds size and type of the animal can influence whether a civilian will
submit a service request for its removal. Outside the realm of road kill, civilians also may submit
a service request for removal of a dead pet in order to avoid paying the fees associated with
28

proper cremation or burial (Wessel 2016). While the removal of a domestic animal may differ
from the removal of road kill in terms of emotional attachment, the public mechanism for
removal disposes of it in “the same bureaucratic dehumanized fashion” (Lulka 2013).
1.5.6. Broken Streetlights
Street lighting contributes to the built environment and context of a location. Depending
on the nature of implementation, lights impact and determine the “aesthetics, cleanliness, traffic
and crime safety and community support or cohesion” of an area (Williams 2007). Residents or
visitors of an area will spend more time outside if they feel the area is conducive to activity and
has higher walkability characteristics, characterized by lighting. In this regard, street lights act as
social support features that promote or deter activity in the built environment.
Lighting is rapidly being considered in a shift towards situational crime prevention where
environmental design improvements restrict opportunities for crime and suppress fear-invoking
environments (Pain 2006). Lighting and visibility are subconsciously associated with an
increased sense of safety. Separate from feelings relating to personal safety, an increase in
“community pride and sense of ownership of the local area” may also drive people to report
broken streetlights (Pain 2006). An online platform for fixing broken street lights in France
called Signalez-nous saw continual use and 311 reporting still occurring after five years of
introducing the application; potentially providing evidence for the belief that the impact of
environmental determinants has a direct effect on citizen’s physical and psychological health,
which is why reports occur (Composto 2016). It is worth noting that for studies focusing on
streetlights and crime that the presence of working street lights may not have actually reduced
crime, but only reduced the fear of crime occurring (Pain 2006).
29

1.6. MyLA311 as Starting Point for City’s Open Data
Los Angeles Mayor Eric Garcetti signed a 2013 Executive Directive designed to promote
transparency and accountability by providing raw data in easy-to-find and accessible formats. In
addition to a transparency effort, the directive aimed to modernize city government and
operations management as a policy goal. The mayor emphasized how a significant goal of the
directive was to “foster creative new thinking about solving our most intractable challenges
through public-private partnerships and promoting a culture of data sharing between our own
City departments and other civic resources” (Office of Los Angeles Mayor Eric Garcetti 2013).
The city has been slowly building up its data and technology resources in an effort to set
a new standard for cities intersecting data and policy. The shift towards open data sought to go
beyond simply sharing information but providing opportunities for the data to transform into
tools with tangible uses for constituents. An Open Data Guide for the city specified that only
data which “increases public knowledge about department operations, furthers the mission of the
department, creates economic opportunity, or responds to a need for public information” should
be placed in the open data repository (Currie 2016). While the city does make some data public,
much of the city’s data is only accessible internally to city employees; for example, the majority
of the service requests displayed previously in Table 1 are not available through the Open Data
Portal. While including MyLA311 data is a start to truly achieving transparency and accessibility
of city data, Los Angeles still must still make strides towards providing more city produced data
to the public.
In 2016 the City of Los Angeles entered another contract to develop an open repository
for the city’s geospatial data with Esri, a GIS company known for creating powerful mapping &
spatial analytics software, for an open data website called the GeoHub. The GeoHub offers the
30

capacity to provide file types conducive specifically to GIS work and was the source of the city
boundary data used in this thesis. According to Esri, the partnership had the goal of making each
department's data “available online in real time (or near real time) to boost efficiency and
eliminate the information bottleneck” (Esri 2016). Much of the initial public spatial datasets
shared came from existing city data repositories, such as MyLA311, with the purpose of
increasing citizen engagement and analysis of city data. Sharing data can lead to democratic
action within a community and overcomes bureaucratic obstacles while providing increased
levels of transparency into city services and operations.
1.7. Thesis Structure
This thesis is comprised of five chapters including this introductory chapter. Chapter Two
provides a comprehensive literature review examining research related to volunteered geographic
data, 311 systems, and reasons for reporting service requests. Chapter Three details the data and
the methodology used analysis following a case study detailing the relationship between service
requests and demographics as they relate to multiple cluster analysis methodologies. Chapter
Four details the results of the analysis with maps, charts, and graphs of the results of multivariate
clustering. Chapter Five interprets the multivariate clustering results of the 311 data and
discusses the benefits from an analysis using MyLA311 data. Chapter Five also includes a
broader interpretation of the overall results of this thesis project.

31

Chapter 2 Literature Review
To better understand the relevance and significance of the spatial analysis of 311 data, it is
imperative to provide background information and discuss previous literature on the topic. The
chapter begins with a discussion of VGI and how MyLA311 data is related, following the shift
towards modernizing civic operations and embracing open data. Next, the 311 data and service
requests are explored more regarding GIS analyses and the background behind certain service
request types. The chapter ends with a brief examination of characteristics related to those who
request 311.
2.1. Volunteered Geographic Information
VGI is geographic data provided voluntarily by individuals or crowdsourced from user
generated content (Goodchild 2007; Sangiambut 2016). The growth of VGI is discussed in this
section, followed by the utility and ease of acquisition of user-generated data representative of an
area. Finally, as the data is created by amateurs, the spatial data quality of VGI is discussed in
relation to submitting service requests.
2.1.1. Ease of Acquisition and Collection of Data
Acquisition of data informs and drives GIS work. VGI offers an alternative source for
data that is both cost-effective and allows for community-driven input in a timely manner. VGI
can engage the public while acknowledging a low barrier for entry regarding capital and
expertise. Advances in technology of both accessible mobile devices and web-based applications
helped to popularize mechanisms for VGI data collection (Sangiambut 2016). The reduced cost
and increased accuracy of GPS receivers alongside an abundance of wireless networks propels a
multitude of readily obtained geographic information (Lu 2016). Previously, mass information
32

from individuals have been collected in more traditional and timely methods, such as through
“telephone, fax, email, and direct meetings with the local administration or the mass media,
which are often time-consuming and unsuccessful” (Brovelli 2015). The ease of both collecting
and supporting VGI opens doors for new types of research and provision of spatial data while
providing insight to scientific and policy-based research.
A VGI system has a directly participatory component to it, as users are generally aware
that shared information contains a locational component. Many applications on devices take into
account user information, with location as a key descriptor of a user (Whitaker 1980). Collection
of this VGI can come from users putting georeferenced information or data in publicly accessible
spaces, such as social media posts, or applications that take into account a location in the
backend of an interface. Data made publicly available can then be freely downloaded as crowd-
sourced information, even if users are not notified about their posts’ information being
downloaded for use by someone else. In joining a platform for social media and agreeing to a
certain set of privacy settings, users legally, and sometimes indirectly, provide consent to sharing
their produced VGI from the medium they shared information on. A more active, user-focused
version of VGI entails a user running applications that gather multimedia data from device
sensors, georeference the location and publicly disseminate it over the Internet (Brovelli 2015).
Ease of sharing geographic information or data leads to more available data, therefore providing
the possibility of more insight. This is not to say that all VGI is high quality, however, as it
places responsibility of data quality on the users; increased quantity and assured quality of data
leads to greater benefit.
33

2.1.2. Spatial Data Quality
The quality of VGI is dependent upon the context of users or source of the information.
In 1980, the US Federal Government promulgated that a geospatial data standard must adhere to
the five dimensions of positional accuracy, attribute accuracy, logical consistency, completeness
and lineage (Goodchild 2012). As GIS technology advanced, a revised set of data quality
standards sought to emphasize completeness, logical consistency, positional accuracy, temporal
quality, thematic accuracy and usability (Antoniou 2015). These characteristics were chosen to
ensure that all GIS data produced, including VGI, can measure the totality of features, provide
adequate background to the context of data collection, entails a certain degree of details and has
no discrepancy between the actual attributes of real-life occurrences and the respective coded
attributes in the data. Metadata from VGI can be designed and combined to garner information
about the users creating the data in an effort to attribute ownership and assess the data quality
(Fonte 2015). The party behind the interface processing the data can design the interface in such
a way to prioritize and provide assurance of higher quality data. As a result, those seeking VGI
integrate design parameters adhering to the spatial data quality standards in order to produce the
best product (Goodchild 2012).
Through contributions to VGI, people drive the collection of data from the masses. More
participation ultimately means a larger sample of the population is able to contribute, providing
more information to accurately describe a space. VGI replaces costly or time consuming methods
for crowd-sourcing data through mass-produced methods at the individual level. Additionally,
crowd-sourcing data entails referring the data to people without respect to their qualifications
(Goodchild 2012). The data must be analyzed for participation patterns of contributors and on
the content created to identify preferences and participation bias within a study (Antoniou 2015).
34

Knowledge of a community and the human component behind making contributions to VGI aids
in understanding the level of different data quality elements (Antoniou 2015).
Quality can differ depending upon individual preferences from user input. As a collection
of sources, a VGI dataset is greatly influenced by “spatial preferences, feature type preferences,
and mapping behaviors” (Bégin et al. 2013). In the context of this research, an individual’s
spatial feature type preferences may determine an inclination to only submit service reports
regarding homeless encampments or graffiti removal yet abstain from reporting illegal dumping
or broken street lights. Citizens inherently, and sometimes unknowingly, have different ranking
levels of prioritization in creation of VGI. These priorities may change dependent on the
proximity of a problem from a place a user spends time or how strongly a user feels about
solving the problem; as a result, VGI is subjective with regard to the user (Bégin et al. 2013; Lu
2016). While the 311 service requests cannot capture every occurrence of the aforementioned
problems in each census tract, the quantity of data amassed over a year within a coherent, high
quality data format can provide insight that could not be found otherwise through analysis.
In the case of 311 data, the service requests are a VGI byproduct collected under the
assumption that people act directly to report factual problems to the city in order to alleviate
them. While more active forms of VGI, such as OpenStreetMap, have a user base with the
intention of accurately mapping and providing information, more passive forms resulting in VGI
as byproduct of users may not have the same spatial literacy (Neis 2012). Analysis of community
characteristics for reporting requires the sociocultural context for individual contributions to VGI
repositories in Los Angeles to be observed. City service requests do not take an individual’s
requirements for credible information into account or provide commentary on the background of
the reporter. Understanding the amount and type of service requests coming from an area and
35

incorporating sociodemographic information alongside 311 data analysis can provide clues to the
connection between service requests and the characteristics of communities reporting them.
VGI raises implications for spatial data quality, including the credibility of the source and
the accuracy of the attributes and positional information (Goodchild 2011). The motivation for
supplying VGI data can suggest a more or less potential for bias or deception, resulting in
implications regarding credibility. Flanagin (2008) seeks to determine the credibility of VGI and
how suppliers in a market should address the sensitivity of VGI users to inaccuracies and
misperceptions. Generally, the context of the source of VGI must be taken into account. As GIS
data shifts to more user-generated content, the potential for politicization or manipulation of the
data is ever more present. Research and analysis involving VGI must explicitly state the source
of VGI and address any possible bias to avoid arriving at false conclusions of VGI data.
Resulting spatial analysis combined with an understanding of the context of VGI’s metadata and
sourcing is then more likely to provide credible information and results.
With the prevalence of mobile devices and increasing accessibility of cloud applications,
such as through applications like MyLA311, every citizen has the potential to be a sensor,
creating data about their environments (Goodchild 2007). Data is now accumulating faster than
ever before as a result of technological advances. It begs asking what drives people to provide
data, how accurate is the data, what the threats towards individual privacy are, or which
information providers are trustworthy (Flanagin 2008; Goodchild 2011). Cities have the
opportunity to provide mutually beneficial means of acquiring this data through 311 services.
VGI takes on the role of that of amateur geographic observation, at times even providing
a comparison for professionally attained geographic information. While VGI may not always
collect every single point of existing data, or in this case every location of a service request, it
36

provides a valuable opportunity on a cost-effective platform to perform geospatial analyses.
Accessibility, either directly through citizen engagement or indirectly through mass data created
by citizens utilizing third-party applications in the process of acquiring and using geographic
information, VGI has the potential to alter GIS landscape significantly and soften criticisms
regarding its use and the quality of produced data (Goodchild 2011).
2.2. Open Data
The VGI provided through the City of Los Angeles MyLA311 website falls into the
category of open data. Open data is publicly available information accessible without having to
make a request to the government (Scruggs 2013). Influenced by President Barack Obama’s
national “Open Government Directive” in 2009, local governments started to provide public
information resources available in “machine-readable” formats (Carrizosa 2013). The US federal
government further emphasized data as a valuable national asset through Obama’s 2013
Executive Order providing a template for data quality and management. Electronic governance
(e-governance) and new forms of civic engagement became popularized through the increased
use of advanced technologies of the government.
2.2.1. Shifting Towards Open Data
Dissemination of government data potentially finds use as a neutral guide for solving the
cities problems and allocation of resources, all while informing citizens. As local governments
gradually adopted new technology, they accumulated more data than ever before; this data
continues to grow at an exponential rate. As constituents became aware of the data’s existence,
they fought for access to it. Consumer rights activists argued information is explicitly linked with
political authority and control, giving way to the Freedom of Information Act (FOIA) declaring
any person has the right to request access to federal agency records or information (Currie 2016).
37

Sharing government data with citizens was initially a slow process due to the lack of channels for
dissemination. Rather than letting the data go untouched and unseen by the majority of a
population unless prompted for, the FOIA and technological openness serve to provide
standardized formats, low costs and ease of access to government data. Citizen access to
information is fueled in part by “openness as government disclosure and openness as open
systems” providing value through entrepreneurship contributing to support of open data policies
and the open data movement (Currie 2016). Newly available open data prompted new ways of
thinking and ideas for approaching existing problems through a fact-based background.
Through accessibility and freedom of restrictions regarding use, the open data model for
managing government-created data is a transformation from prior internal systems. In order to
contribute to a city’s data repository, city records are often digitized in order to provide a
plentiful source of economic and institutional capital capable of propelling data-driven
management and city planning goals in both the private and public sector (Currie 2016). Open
source communities promote activity and content sharing on the end of the user and promote
increased production of data and collection from the side of the publisher (Brovelli 2015). Open
data produced by the government should be non-partisan in existence and removed from the
context of its creation. Over time, accumulated data can be examined for identifying resulting
effects in a community regarding new data mediators, policies or practices (Currie 2016).
2.2.2. Public Participatory GIS and E-governance
Open access and information sharing leads to civic engagement. Many e-government
initiatives, which entail “delivery of information and services online through the internet and
other digital means,” are accessible from personal devices, as opposed to the previous process of
physically going to a government entity for more information (Lu 2016). VGI and public
38

participatory GIS (PPGIS) provide an opportunity for heightened public involvement through
open access tools and accessible/actionable information. VGI focuses on both active and passive
collection of spatial information from citizens as sensors while PPGIS tends to inform
government planning agencies through solely active public involvement (Lu 2016). The
participatory component of VGI through complex modes of engagement can affect the types of
data collected for a decision process (Brown 2013). The term PPGIS emerged in the mid-1990s
for specific use cases where GIS was used with community intervention to aid insight towards
policy decision-making; there was a shift from viewing GIS through only technical lens to a
perspective involving institutions and greater society, all while emphasizing democracy (Brovelli
2015). In either case, individuals have the opportunity to create their own unique data that can
further drive other processes.
Involving citizens more through open access and technology is bringing about new forms
of governance. E-governance uses the internet a platform to bridge each individual to city hall,
providing easily accessible opportunities to interact with the government. E-governance at its
various stages (Figure 13) must be designed to encourage the public to become involve while not
dissuading them through ease of operating and interacting. The stages progress from Stage 1
with one-way communication from the government to constituents with zero interactivity to
Stage 5 with robust, two-way communication that is integrated into all key functions of the
government and advances with the latest technology to best serve citizens.
39

Figure 13 Stages of E-government, source: Moon 2002
Within the context of this analysis, the MyLA311 system falls between Stages 2 and 3 of
e-governance through the focus on customer service fulfilment for citizens and the city.
Involving people in the production of civic data is a process of co-production through e-
governance (Sangiambut 2016). Within this framework, a government no longer solely provides
services but promotes market-based decision making as a manager of service providers (Hood
1995). In this regard, the functions of the government are shaped by how well they use
technology and open source software/interfaces to interact with citizens. The design and user
interface of the technology created for engagement decides the scale of how democratizing effect
data production and use will have on a city (Sangiambut 2016).
Accessibility of interaction between both the user and the service leads to high levels of
engagement for both parties. Open access e-government is no longer a one-way transaction of
40

“government-to-citizen”; rather it opens up a two-way “citizen-to-government-to-citizen”
dialogue and exchange of ideas (Sieber and Johnson 2015). The ability of online interfaces
allows the public to stay regularly informed through the ability to provide updates at a minimal
cost. GIS technology must empower the public through its exploitation to effect policy and
governance (Brovelli 2015). Data produced by individuals can be incorporated into management
and aid decision-making for policy or other action within a community. Accessibility of e-
government lessens the gap between constituents and the government by providing means for
transparency, responsiveness and accountability (Lu 2016).
Citizen involvement in government processes is mutually beneficial, but there must be
action on both ends for a dynamic understanding of the urban environment. The rescripting of
citizen engagement should not place a burden solely on citizens to be active agents in claiming
and monitoring their urban spaces (Sangiambut 2016). Participation translates into existing
citizen participation from the real world to the digital world within the context of contributing to
GIS data (Brown, 2013). As cities modernize and integrate more data sharing capabilities into
their operations, individuals are able to extol individual emancipation from inefficiency
(Sangiambut 2016). Coproduction of data as a byproduct of smart government initiatives allows
cities to find opportunities for growth in understanding of technology, organizational
management and how to approach challenges to better meet the needs of constituents (Schwester
2009).
More recently and relevant to the study area, Currie (2016) details Los Angeles’s
relationship with open data and how it seeks to “enable new administrative models and inspires
new modes of civic involvement” on multiple fronts and mediums, such as the Facebook post in
Figure 14 posted by L.A. City Councilwoman Monica Rodriguez’s office. Through online
41

marketing citizens are more likely to engage with a civic entity, in this case a city councilwoman,
and have it feel akin to normalized interaction outside the realm of the civic environment.
Currie’s research details how open data reshapes modes of administration and policy through a
modernized, data-centric lens appealing to both citizens and the government. As a result, analysis
and research has been performed on different types of existing 311 data, providing a baseline for
how to utilize 311 data while acknowledging inherent limitations of the data, as discussed in a
later chapter.

Figure 14 A Facebook social media post encouraging self-reporting by constituents
2.3. Analyzing 311 Service Requests
As descriptions of events and of spatiotemporal attributes, service request data is
naturally tailored for use in GIS analyses revealing patterns relating to human nature. Spatial
analyses of 311 service requests require a vast amount of data to provide a baseline where
patterns of spatial distribution can be identified and avoid spatial autocorrelation (Mullen et al.
2014). Service request data can be seen as representative of the state of the “built environment,”
42

comprising of urban design, land use and the transportation system to encompass patterns of
human activity within the physical environment (Gehl 2010). Changes to the built environment
exist on a broad time spectrum, with some fleeting and others remaining for longer periods of
time to shape the lives of people. The spatiotemporal fidelity through an address and the creation
and service dates of 311 service requests types allows for in-depth analyses spanning a multitude
of fields.
The human component of submitting a service request can be tied to a human-centric
analysis of the environment. O’Brien (2016) finds that reporting of civic problems is
distinguished behaviorally from reporting public issues arising from natural deterioration and
people are found to specialize in one or the other. Investigating the type of person who self-
reports community problems may identify that basing community intervention solely based on
311 data perpetuates inequalities simply because some communities may be partial to dealing
with problems in their own way and not through self-reporting.
In a later publication on 311 data analysis, O’Brien (2017) finds the coproduction of data
translates civic motivations into impacts. Many of these motivations are not themselves
inherently civic, yet the submitted data platforms channel reports into positive outcomes for the
community. Kontokosta (2017) finds that neighborhoods under reporting 311 data are often
characterized by higher proportions of males, unmarried individuals and minorities, alongside
higher unemployment and a smaller population of individuals fluent in English, providing further
evidence for socioeconomic status and household characteristics have a non-trivial effect on the
propensity to submit 311 service requests.
When analyzing 311 service requests of an area, it is important perform the analysis at a
scale that will reveal useful patterns and new information at the local scale. Drawing on previous
43

research on citizen–government interaction, service delivery, and civic engagement, Minkoff
(2015) focuses on how contacting propensity and condition both explain spatial variations in
service request volume. Through an exploration at the census-tract-level variation in 311 service
request volume within New York City, he finds using a larger administrative dataset at micro
places allows one to estimate a more precise effect than audits at a smaller number of places
would. The perspective of 311 occurring at an increasingly localized level is emphasized through
White (2016) ’s evaluation of 311 data as a measure of neighborhood-level realized demand for
services and provider of information regarding relative intensity of how local neighborhoods
utilize government services rather than 311 data as a measure of political participation and
propensity for participation. Similar to the methods used in this analysis, she uses 311 data as a
proxy for neighborhood-level propensity to participate and noted limitations. White (2016)’s
research provides examples of how to best visualize the data both geographically and
numerically while not committing an ecological fallacy.
2.4. Multivariate Clustering
Service request data sets comprising of more than a few thousand objects benefit from the
assistance of algorithms for identification of patterns present in the data. Ester et al. (1996)
defines a clustering algorithm as a process used for class identification to group a database’s
objects into subclasses with intentional meaning. The density-based notion of clustering lends
weight to the idea of similar clusters comprising a neighborhood (Ester et al. 1996). Performing a
multivariate clustering analysis of service request type frequencies located in a defined space, a
census tract, provides the points needed for identifying distinct signatures in cluster types.
Within ArcGIS Pro, a multivariate clustering analysis establishes natural clusters of feature
characteristics based solely on feature attribute values within polygons.
44

Seeking to understand unique signatures within urban context at a localized scale, Wang
et al. (2017) uses R to provide a multivariate cluster analysis of three cities’ 311 data identifying
spatial signatures of parts of the city that have similar patterns of requesting services.
Sociodemographic patterns are identified within the clusters established from the multivariate
clustering analysis of 311 service request type patterns’ locational attributes through census data.
The analysis produced a methodology for creating a low-cost decision support tool for urban
stakeholders seeking information at the local level from city service data. The research warranted
that 311 data analysis models based on regularly-updated open 311 data can have considerable
potential regarding insight of city operations and neighborhood planning; it remains noteworthy
that the structure of 311 reports correlated with socioeconomic quantities does not serve as an
evidence of any causal relation, yet still provides insight to conditions present in the built
environment (Wang et al. 2017).
Wang et al. (2017)’s multivariate clustering methodology was informed by previous
analysis revealing the different structures or characteristics within cities. Bettencourt (2010)
clusters cities into different classes of urban dynamics through analysis of data for gross
metropolitan product, personal income, violent crime and patents for the construction of
meaningful, science-based metrics for ranking and assessing local features. Spatiotemporal
assessment of the characteristics of the cities leads to advancement in understanding the theory
of urban evolution and promotes new tools for the formulation of improved urban policy.
Kontokosta (2017) produces a proof-of-concept model of analyzing New York’s 311 service
requests to create a benchmark for and validate neighborhood resilience capacity. Kontokosta
(2017) determines the average levels of community resiliency and activity at the neighborhood
45

scale through analysis of five years’ worth of 311 requests for service in New York as a proxy
for local activity.
In a multivariate cluster analysis, the parameters determine the course of the machine
learning analysis. In Wang et al. (2017)’s analysis in R, the Initialization Method parameter uses
Random seed locations option to frame the clustering algorithm as a sensitivity analysis to
identify which features are always found within the same cluster through randomly selected seed
features. One drawback of Random seed locations is that the algorithm incorporates heuristics
and can return differentiating results each time the tool is ran, despite identical parameters and
data. Two options available for clustering algorithms are K-means and K-medoids. The K-means
function, described in Figure 15, uses machine learning and iterative refinement between
classifying the characteristics making up a cluster and then computing the mean/centroid of all
data points within the cluster type until there is little change within the makeup of the clusters
(Trevino 2016). While K-medoids is more flexible in accommodating noise and outliers in the
input features, K-means is faster and preferred for larger datasets, often bolstering the decision to
use it in an analysis (Esri 2018).

Figure 15 K-Means Cluster Function (Sayad 2019)
The multivariate clustering tool often has the option to generate the optimal number of
clusters, yet Wang et al. (2017) notes that four clusters is the best overall metric, as demonstrated
through a comparison of results attained using both the Silhouette method and the Elbow
46

method. The Silhouette method is a commonly used method of interpretation and validation of
consistency within clusters of data. It measures how similar an object is to its own cluster
(internal relation) with other clusters (external relation). The Elbow method observes the
percentage of variance explained as a function of the number of clusters using a pre-determined
set of information. In summary, Wang et al. (2017)’s research found that within use of both
Silhouette and Elbow methods, only four clusters were presented in analysis of three separate
cities (New York, Chicago and Boston), as this number of clusters revealed detail about a city’s
sociodemographic profile while providing a reasonable trade-off between having too many
clusters and an acceptable clustering quality.
2.5. Collective Efficacy
Collective efficacy relates to specific tasks taken to maintain public order. The idea of
collective efficacy revolves around the engagement of community members through a basis of
trust and a willingness to intervene for the common good to effectively create a safe and orderly
environment. Collective efficacy can help stratify the social characteristics of a place of
residence or work through inherent structural contexts (Sampson 1997). While much of the
research surrounding collective efficacy is done within the context of crime, the underlying
connection of the residents taking action to remedy physical ailments of their community
prevails. Areas with similar attributes form clusters of individuals with common values as
residents to maintain effective social controls with variation depending on the neighborhood
(Sampson 1997). The collective actions of individuals inhabiting an area then enact change
through a shared mindset.
The action of submitting a 311 service request is aligns to a citizen exercising informal
social control to better influence the community and shape public order to a certain set of
47

standards. A service request is a task-specific construct built upon citizen engagement and shared
expectations (Morenoff 2001). In submitting a request to remedy a physical malady of the
environment, people become transformational agents heavily influencing environments around
them. Participation in the collection of data can bring about many benefits and provides an
opportunity for collaboration on achieving social, economic, political, scientific and
environmental goals (Lu 2016). In effect, it is their individual beliefs of personal efficacy that
motivate and guide their actions, further influencing those around them. This shared sense of
collective efficacy through collective influence lends citizens success in shaping and bearing
command on their surroundings (Bandura 2000).
Sometimes, people in a community can instead place more weight on the disorder present
in the built environment negatively influencing the actions of the people who reside in it rather
than focusing on the altruistic act of people bettering the built environment to make it more
preferable for living. While there is still an emphasis on betterment of the community as an end
goal, the “collaborative effort to maintain a certain standard of communal life” stems from
negative connotations (Wilson 1982). Through this perspective, small problems in the
environment can easily result in much larger problems occurring down the line. This reasoning
purely stems solely from the physical component of disorder: each service request included in
the analysis representing an attempt at self-policing a physical problem in a community.
If community members do not report a service request for an ongoing problem, it is
inferred that the problem is engrained within the fabric of the community and attempting to
remedy it would be considered feckless. However, if a service request is seen as an attempt to
raise the standard of living throughout the community if it is made with the intention of
remedying the problem and providing the requestee with a level of social control. Within the
48

scope of research by Wilson (1982), a lack of cohesion in a community reduces chances of any
one person acting as a representative agent of the entire community. While this view of
collective efficacy is more relaxed, the mindset does much more for simply (temporarily)
remedying a problem, rather than attacking the problem at the root.
The success of the MyLA311 Program relies heavily upon citizen engagement. Citizens
are the “coproducers” of public services through interactions with service agents to redirect city
services. Rather than an agent presenting a finished product or service to the citizen, the agent
and citizen produce the desired transformation together to effectively exert influence on both
policy and maintenance of city programs (Whitaker 1980). The number of requests
disseminating from an area can lead to revisions in the management, supervision, and training
regarding the request type, therefore affecting the distribution of service allocations within the
city. City management can then better predict the need for specific service requests in different
locations and distribute the funds and resources appropriately.
In a city as large as Los Angeles, citizens take pride in the upkeep of their communities,
referring to certain areas as distinct neighborhoods to gain a more geographically distinct
identity. The city must actively respond to the changing tastes, circumstances and behaviors of
constituents and their request for assistance in tackling new problems (Whitaker 1980). Through
the data, it is evident that service requests created by city employed agents make up a sizable
portion of the requests, it is clear that city services are not likely to be as helpful as they could
and should be without the citizens themselves taking action. Collective efficacy, in all of its
forms, is rooted in community action with long-lasting benefits, all of which stem from citizens
taking the initiative towards bettering their surroundings.
49

Submitting a 311 service request provides a citizen with the opportunity to influence their
surroundings and physical aesthetic of a location while exerting some sort of social control.
Within this research, there is a focus on the aesthetic of a location due to the physical attributes
that the chosen service requests impact. This strong “sense of place,” or locational identity,
combined with the demographics of an area shape its local signature (Handy 2002). Determining
the reasoning behind civic engagement of service requests can be supported through the theories
behind collective efficacy.

50

Chapter 3 Data and Methodology
This chapter describes the data and methods used to evaluate the relationship between MyLA311
service request data, census data, and the locations of service requests. First, this chapter
describes the data used in the analysis. The first phase of this analysis involved identifying the
types of service requests used and the resulting temporal scale. Next, the census data and its
attributes are discussed, as well as other geographic boundaries used in the analysis. Finally, the
methodology behind how and why clustering is used is discussed.
3.1. Methods Overview
The methodology for this thesis demonstrates the process of the steps taken for an
analysis of the MyLA311 data (Figure 16). The data acquisition entails using publicly available
data to form the basis for the analysis. Data preparation involved cleaning extraneous fields and
removing irrelevant service requests in both Excel and ArcGIS Pro to ensure only relevant data
was taken into consideration. This thesis uses ArcGIS Pro’s machine learning geoprocessing
Multivariate Clustering tool to determine clusters with similar features regarding 311 reporting
frequencies. The tables produced from the multivariate clustering analysis tool were then brought
into Excel for further analysis and visualizations through radar and bar charts.
51

Figure 16 Overview of methodology process
3.2. Data and Processing
All of the data used in this analysis is publicly available, allowing the analysis to be
reproducible. All data from MyLA311 used in the analysis is from 2018. While the source of the
data is MyLA311, the data can be found and downloaded through Data.LACity.org and is
provided by the City of Los Angeles Information Technology Agency in the category “A Well
Run City”. Shapefiles for the city boundaries and other administrative geographies can be found
through the Los Angeles GeoHub. Finally, the 2017 5-Year Estimate American Community
Survey (ACS) information can be found through the United States Census Bureau website.

52

Table 2 Data sets used in the analysis
Name Spatial Unit Source Data Purpose
MyLA311
2018 Data
Latitude and
longitude of points
Los Angeles
Open Data
Portal
Illegal dumping, graffiti
removal, homeless
encampments, dead animal
removal, broken street lights
2017 5-Year
Estimate
ACS Data
Census Tract level
for City of LA
U.S. Census
Bureau
Determines demographic /
socioeconomic characteristic
traits in relation to service
requests
TIGER/Line
Census
Tracts
Census Tract level
U.S. Census
Bureau
Will be used to spatially join
data. Offers visual context of
data and boundaries
3.2.1. MyLA311
The MyLA311 data is comprised of service requests made during 2018 for the selected
service request types. As a proof of concept for determining clusters, the scope of the analysis
required only one year of service requests, yet the temporal and spatial component can be made
larger or more granular depending upon the goals of an analysis. The frequencies of the service
requests served as exploratory variables later on as the basis of the multivariate clustering.
MyLA311 data was cleaned of any personal identification information and extraneous
commentary or attachments and then reproduced by the City of Los Angeles My311 Data
Management team. The origin of the data stems from individual requests for service made to the
City of Los Angeles. When the service requests are geocoded into individual points on a map
through respective latitude and longitude, each point falls within a census tract polygon. Figure
17 is representative of the MyLA311 service request points at their base level on a map
symbolized by service request type.
53

Figure 17 MyLA311 2018 data displayed as points at a large spatial scale in a section of South
Los Angeles
Several steps needed to be completed for quality control of the point data. The
downloaded data was comprised of service requests for only illegal dumping, graffiti removal,
homeless encampments, dead animal removal, and broken streetlights. These service requests
were chosen because of their visual characteristics, which prompt individuals to report them. The
raw data contained many attribute fields, exemplified in Table 3, and had to be cleaned through
filtering and removing attributes not essential to the analysis, which mainly used SR Number,
RequestType and location features. Additionally, city employees completing requests on the job
have the option to self-report the need for a service request when they are out in the field in order
54

to provide a record of addressing an issue. These service requests, along with any with a
“cancelled” status, were filtered and then removed from the data set in Excel so that the analysis
solely focused on reports made by citizens.
Table 3 MyLA311 Data Attributes

Within this analysis, data for broken streetlight requests was comprised of two separate
requests for single streetlight repairs and requests for streets with multiple broken streetlights. In
order to ensure that the data was within the City of Los Angeles boundaries, data was queried to
only include service requests with “Y” in the AddressVerified attribute, therefore ensuring that it
was internally validated by the city with GIS data. This validation comes from geocoding the
given address of a service request through the city’s internal Thomas Brothers Map locator with
all addresses within the city. While each service request attribute includes an address, the
supplied latitude and longitude attributes generated by the city were used in order to avoid
manually geocoding any of the addresses, which would have been a lengthy and time-consuming
process.
3.2.2. American Community Survey Data
The census data stems from the 2017 5-Year Estimate ACS and contains both
socioeconomic and demographic variables at a variety of spatial aggregations. The data provides
insight regarding population diversity, education, income and employment at the scale of census
55

tracts. All of the data is freely available through the United States Census Bureau’s American
Fact Finder Download Center. The data was processed, as described below, to only include
attributes needed in the analysis.
Following the methodology used by Wang et al. (2017), demographic features represent a
variety in population diversity, education, and income and employment in this analysis. In the
case of the data from the ACS, data was downloaded with ‘Census tract’ selected as the
Geographic Type in the ‘Geography Filter Options’. Additionally, the selected data contains
information from the following categories: “Hispanic”, “Non-Hispanic White”, “Non-Hispanic
African-American”, “Non-Hispanic Asian”, “High school degree”, “College degree”, “Graduate
degree”, “Uninsured ratio”, “Unemployment ratio”, “Poverty ratio”, and mean for “Income
(all)”, “Income of No Family”, “Income of Families” and “Income of Households.” The datasets
created by the U.S. Census Bureau in the American Community Survey used in the analysis are
detailed in Table 4.
56

Table 4 American Community Survey Census Datasets
Census Dataset GEO.display-
label
Attribute Fields
ACS_17_5YR_DP05 HC03_VC93 Percent; HISPANIC OR LATINO AND RACE -
Total population - Hispanic or Latino (of any race)
ACS_17_5YR_DP05 HC03_VC99 Percent; HISPANIC OR LATINO AND RACE -
Total population - Not Hispanic or Latino - White
alone
ACS_17_5YR_DP05 HC03_VC100 Percent; HISPANIC OR LATINO AND RACE -
Total population - Not Hispanic or Latino - Black or
African American alone
ACS_17_5YR_DP05 HC03_VC102 Percent; HISPANIC OR LATINO AND RACE -
Total population - Not Hispanic or Latino - Asian
alone
ACS_17_5YR_S1501 HC02_EST_VC11 Percent; Estimate; Population 25 years and over -
High school graduate (includes equivalency)
ACS_17_5YR_S1501 HC02_EST_VC13 Percent; Estimate; Population 25 years and over -
Associate's degree
ACS_17_5YR_S1501 HC02_EST_VC14 Percent; Estimate; Population 25 years and over -
Bachelor's degree
ACS_17_5YR_S1501 HC02_EST_VC15 Percent; Estimate; Population 25 years and over -
Graduate or professional degree
ACS_17_5YR_S1701 HC03_EST_VC01 Percent below poverty level; Estimate; Population for
whom poverty status is determined
ACS_17_5YR_S1903 HC03_EST_VC02 Median income (dollars); Estimate; Households
ACS_17_5YR_S1903 HC03_EST_VC22 Median income (dollars); Estimate; FAMILIES -
Families
ACS_17_5YR_S1903 HC03_EST_VC47 Median income (dollars); Estimate; NONFAMILY
HOUSEHOLDS - Nonfamily households
ACS_17_5YR_S1902 HC03_EST_VC02 Mean income (dollars); Estimate; All households
ACS_17_5YR_S2301 HC04_EST_VC01 Unemployment rate; Estimate; Pop. 16 years and over
ACS_17_5YR_S2701 HC05_EST_VC01 Percent Uninsured; Estimate; Civilian
noninstitutionalized population

In order for the data and analysis to accurately portray the demographics of Los Angeles,
it is important to carefully address the concepts of race and ethnicity. Race and ethnicity are
categorized separately in the census, and respondents may report any combination of race and
ethnicity. For the 2017 ACS, the census allowed individuals to report race as one or more of the
following categories: American Indian and Alaska Native, Asian, Black or African American,
Native Hawaiian and Other Pacific Islander, White or some other race. Ethnicity was broken into
two categories, “Hispanic or Latino” and “Not Hispanic or Latino.”
57

For this analysis, the categories of race and ethnicity were combined. Four categories of
race and ethnicity were identified:
1. “Hispanic,” which included all ethnicity responses of “Hispanic or Latino” and any
selection for race
2. “Non-Hispanic African-American,” which included responses for ethnicity of “Not
Hispanic or Latino” and race of “Black or African American”
3. “Non-Hispanic Asian,” which included responses for ethnicity of “Not Hispanic or
Latino” and race of “Asian”
4. “Non-Hispanic White,” which included responses for ethnicity of “Not Hispanic or
Latino” and race of “White”
This decision was influenced by the “Los Angeles County: Predominant Racial or Ethnic
Group by Census Tract” map produced by Allen and Turner (2002). The categorizations chosen
for Allen and Turner (2002)’s map accurately encapsulate the diverse demographics of Los
Angeles. Allen and Turner (2002)’s perspective acknowledges the information and context lost
within Los Angeles if only race is symbolized, instead of race alongside ethnicity. The Hispanic
population within Los Angeles is steadily growing and necessary for providing context of the
makeup of the city, as evidenced by the maps procured in Figure 2 in Chapter 1. A difference of
this analysis was that it did not combine “American Indian or Alaska Native” or “Native
Hawaiian or Other Pacific Islander” with the Asian identifying respondents because of a minute
response from those two groups which would not have greatly impacted the outcome of the
census tract clustering.
58

3.2.3. Contextual Boundary Data
Administrative polygon and line shapefiles were downloaded to provide geographic
constraints and context in the analysis. Both of sources used to acquire this data were from
websites serving as public platforms to freely explore, visualize, and download location-based
Open Data in GIS files. The boundary data was downloaded from the Los Angeles GeoHub. In
order to be published on the Los Angeles GeoHub, Los Angeles Information Technology
Agency require the data to have adequate metadata regarding the source and accurate or
precision. The shapefiles served the purpose of providing locational context for the city’s extent.
The TIGER/Line shapefiles were downloaded from the U.S. Census website. Census tract
boundaries were downloaded for Los Angeles County and were clipped to only include census
tracts within City of Los Angeles boundaries. The census tract TIGER/Line shapefiles were the
main polygons used within the analysis, as both the MyLA311 data and sociodemographic data
are joined to the polygons. The same polygons were then used for the frequency analysis of
MyLA311 data points to determine the number of points and types of points in each census tract
polygon. These polygons displayed the cluster signature type in the final maps.
3.3. Data Aggregation
Performing a multivariate clustering analysis requires the points from individual service
requests to be counted within some sort of administrative boundary, in this case polygons
comprised at the scale of census tracts. Given the differing values present (i.e. frequency of
service request types, dollars, percentages), the data must be standardized prior to analysis.
3.3.1. Scale of Analysis
The decision to use census tracts as the level of spatial aggregation for the analysis stems
from a necessity rooted in the need for identifying patterns at the localized level. Zip codes,
59

council districts, neighborhoods, census tracts and census block groups were all considered as
options for data aggregation, yet census tracts provide the best trade-off between spatial
granularity of having a sufficient number of sub-areas within each city and containing a
statistically significant sample of service requests (Wang et al. 2017).
Within the analysis, 1059 census tracts were analyzed. While Los Angeles data originally
had 1170 census tracts, this included sliver polygons from the TIGER/Line census tract shapefile
that may have occurred during the clipping process from slight misalignments of boundary lines
downloaded from the LA GeoHub. Located adjacent to normal polygons within the Los Angeles
city boundaries, the sliver polygons are long, elongated areas which do not represent an entity
and must be deleted. The misalignment of boundary lines and creation of sliver polygons during
the clipping process did not affect the outcome of the analysis. The sliver polygons were
identified through sorting the frequency analysis by count and individually checking the location
of census tract ID’s with zero service requests present. Excess census tracts were then removed
and the spatial join of points within the census tract was finalized to only account for points
within census tracts of city boundaries.
3.3.2. Data Normalization
American Community Survey (ACS) data required normalization for the comparison of
different features. In creating a radar chart, the standardization process was necessary for a
comparison within a population based upon features with different values and units. The units of
attribute values differed based upon the measured feature, such as income in dollars versus a
percentage of people in a population belonging to a category. ArcGIS Pro automatically
standardized data by z-score to allow for proper comparison across unit types through the
multivariate clustering geoprocessing tool and creation of the box plots in Chapter 4. The
60

equations in Table 5 were used to create standardized radar charts in Excel. Within the radar
chart, the data was grouped by attribute type in order to better visually identify differences in
patterns in the resulting chart.
Table 5 Standardizing in Excel
Excel Functions for Standardizing Data
AVERAGE Returns the average of its arguments
STDEV.P Calculates standard deviation based on the entire population
STANDARDIZE Returns a normalized value
3.4. Data Analysis
In order to perform the spatial analysis, all spatial data had to be cleaned within ArcGIS
and Excel for quality assurance (Table 6). This entailed removing data outside of the study area
and irrelevant data prior to performing spatial statistics through the frequency analysis tool and
using the multivariate clustering analysis tool. Tables created from ArcGIS analyses were
exported to Excel and then joined through V-Lookup functions. Pivot Tables and charts were
used to then visualize the final results.
61

Table 6 Software required for analysis
Software Manufacturer Function
ArcGIS Pro 2.3.1 Esri Geoprocessing Functions
• Spatial Join
• Table analysis and management
• Statistical Analysis
• Selecting and Extracting data
• Frequency Analysis
• Multivariate Clustering
Excel 16.23 Microsoft Data Manipulation and Analysis, Charts
3.4.1. Processing and Joining Data to Shapefiles
In its raw format, the MyLA311 and census data downloaded were not embedded into
shapefiles. After downloading the chosen census variables, the data was opened in Excel to be
properly formatted and removed of extraneous information. This process entailed selecting only
necessary columns of data from the various census datasets downloaded for the analysis. Once
all of the columns of data were selected, a V-Lookup function in excel ensured that each census
tract number and its respective values are returned in the final, cleaned Excel sheet to be joined
to the spatial data.
The attributes of the TIGER/Line shapefiles were properly formatted prior to joining
them to census data. The GEOID field was a text field and needed to be a ‘double’ field in order
to join to the ID field in the census Excel data, which was also a ‘double’ field. To do so, the
attribute table of the census tract was opened and a field with a name ‘GEOID10’ and type set as
‘Double” was added. The field calculator then populated ‘GEOID10’ with data from ‘GEOID’.
The census data was then dragged onto the map so it could be joined to the census tract
shapefiles. The ‘GEOID10’ of each dataset was the basis for the matching.
The MyLA311 data was formatted in a way to easily allow GIS use. Once the data was
dragged onto the map, it appeared in the Table of Contents. The data was then able to populate
62

the map by right clicking the table and selecting “display XY data.” The longitude and latitude
fields of the MyLA311 data corresponded to longitude and latitude fields of the join tool,
respectively. After clicking OK, an event layer was created in the map; however, this was just a
temporary layer that must then be made permanent by using the Copy Features tool and
exporting the data to the map. A spatial join then added the census tract values to the MyLA311
points’ attributes.
A frequency analysis was then run to read the attribute table, and it created a new table
containing unique field values and the number of occurrences of each unique field value, thereby
creating a count of each service request type within each census tract. This table was then
exported and manipulated by a pivot table in Excel. The pivot table ensured that each census
tract is in the GEOID10 column and the count of each service request type was in successive
columns. It was then saved as a CSV file and joined to the original census tract shapefile through
matching the GEOID10 values.
3.4.2. Multivariate Clustering Analysis
ArcGIS Pro has a multivariate clustering tool in its spatial statistics geoprocessing
toolbox that uses unsupervised machine learning methods to determine natural clusters of
features based solely on feature attribute values. The classification method is considered
unsupervised because it does not require a set of preclassified features to train the methods used
to find the clusters in the data. The tool works to maximize both within-group similarities and
between-group differences while adjusting for every possible combination of the features to
cluster. While the methodology of the multivariate clustering in this thesis follows the same
parameters of Wang et al. (2017), the analysis was performed using tools available through
ArcGIS Pro.
63

Multivariate clustering analysis was performed only with MyLA311 service request data
frequencies within the census tract polygons. Sociodemographic data was not taken into
consideration regarding the creation of clusters. The first input for the analysis involved selecting
all variables necessary for the multivariate cluster analysis to run. The analysis fields used in this
thesis are the counts of service requests for graffiti removal, homeless encampment, illegal
dumping, dead animal removal and broken streetlights. The counts within these census tracts
served to distinguish clusters of features from one another.
The additional parameters entailed an integer value representing the Number of Clusters
to create, an Initialization Method and clustering algorithm. The analysis was limited to four
clusters with the reasoning detailed in Section 2.4. The random seed locations option for the
Initialization Method parameter was selected to frame the clustering and identify features found
within the same cluster through randomly selected seed features. K-Means, explained in detail in
Figure 15, was selected as the clustering algorithm and was discussed further in Chapter 2. If run
multiple times, the multivariate clustering geoprocessing tool will not repeatedly create the same
output due to the use of random seed locations; however, the created cluster groupings should
generally be similar given the same parameters and inputs. The tool itself is exploratory for this
reason.
3.4.3. Sociodemographic Analysis
In order to analyze the sociodemographic characteristics, the selected data from the ACS
2017 5-Year Estimates were then spatially joined to the new shapefile of census tracts with
resulting cluster types from the multivariate cluster analysis. After running the tool, a new output
layer shapefile was created detailing the cluster type of each census tract. The attribute table of
the new output only included attributes included in the analysis fields parameter from the source
64

layer. For this reason, a spatial join was then run with intersect to add the name of the census
GEOID10 to the resulting cluster type. This new attribute table was then exported to Excel to
create the radar chart of sociodemographic values of the clusters. A VLOOKUP function then
matched the census tract cluster GEOID10 (census tract ID) with the respective
sociodemographic values from the ACS 2017 5-year estimate data set.
Within Excel, the sociodemographic values of each census tract were then separated by
cluster type. The average of cluster values was determined to form a baseline for the
characteristics of a cluster. The average and standard deviation of all census tracts observed in
the analysis was also determined. The data was then standardized by the STANDARDIZE
function in Excel, which calculated a normalized value (z-score) through =STANDARDIZE
([Cluster Average Value], [Average Value of All Census Tracts], [Standard Deviation of All
Census Tracts]). These z-scores were then used for the radar chart in order to visually display the
average characteristics present within each 311-determined census tract cluster at the same scale,
thus allowing comparison of the socioeconomic characteristics of the 311 clusters. The same
methodology was followed to produce radar charts of the service request types per cluster.
Each radar chart displays multivariate results in a two-dimensional graphic by
representing different axes of each exploratory variable starting from the center and extending
outward. Each color-line represents a different cluster type, and the line size drawn connecting
the data to each axis represents the relative magnitude of the variable for the group. The chart
was selected for visually displaying group-observations because the overlay allows visual
comparison of the relative position of each group. Each characteristic was standardized to
achieve an equal weight and to simplify the interpretation by plotting z-scores instead of raw
values on each axis.
65

Chapter 4 Results
This chapter documents the results of the analysis. The Multivariate Clustering Analysis for the
MyLA311 service request data determined the constructed 311 clusters present within each
census tract and the correlating sociodemographic census data. All the exploratory
sociodemographic variables were observed against the clustering metrics to identify statistically
significant correlations. The resulting clusters create urban signatures from 311 frequencies.
Sociodemographic values present in the census tracts of clusters characterize and inform the
features of the local community.
This chapter is broken into several sections to present the results of the analysis. Section
4.1 provides a visualization and a breakdown of the of MyLA311 service request data. Section
4.2 describes the results of the MyLA311 multivariate cluster analysis and 4.3 identifies the
sociodemographic characteristics most present in the clusters. 4.4 summarizes the results of the
analysis within the geographic context of Los Angeles.
4.1. Analysis of Service Requests
Prior to analyzing the patterns present in the MyLA311 data, it was necessary to analyze
the makeup of the data itself within the context of Los Angeles census tracts. Through the
frequency analysis, the counts of each service type and sum of total service requests per tract
were identified. The purpose of the map in Figure 18 is to display where the greatest intensity of
311 calls for service are disseminating from in Los Angeles. The Hollywood Hills, north Valley
and Harbor communities generally have the lowest number of requests. The highest intensity of
MyLA311 service requests are found in the central Valley, Downtown and Westside
communities.
66

Figure 18 Map detailing the amount of 311 requests for service normalized by the standard
deviation of each census tract
67

The overall distribution of service request type and source is detailed in Figure 19. The
majority of all service requests are through the MyLA311 mobile application, phone calls to the
311 hotline or through the MyLA311 website. Figure 19 serves the purpose of showing how
much data stems from the MyLA311 mobile application, mainly due to its ease of use in
reporting a service request. An additional benefit of the mobile application is the ability to add a
photo to document the problem. The multiple mediums available for submitting a service request
are able to appeal to a variety of audiences with varying technological literacy, all while
lowering barriers to accessing a platform or service used for service requests.
Graffiti removal was the most popular service request, followed by illegal dumping
pickups, homeless encampments, dead animal removal, and finally streetlight issues.
Observation of the service request sources revealed the preferences of individuals in regard to
reporting certain service requests. For example, it is more likely a person would prefer to use the
mobile application, followed by the website and phone call to report graffiti removal or homeless
encampments, but would most likely use a phone to report illegal dumping or dead animal
removal before considering reporting through the website or mobile application. Finally, it was
revealed that service requests for streetlights, the smallest group of requests, is evenly spread
across the website, phone, or mobile application as a medium for reporting problems within the
city.
68

Figure 19 Chart detailing the sources of MyLA 2018 service request data and breaking down the
types of the service request types analyzed in multivariate clustering
4.2. Service Request Clusters from MyLA311 Data
This multivariate clustering is done solely on the basis of 311 service request frequency
from within each census tract–sociodemographic values from census data are not taken into
account. Figure 20 is the visualization of service request clustering frequency regarding graffiti
removal, homeless encampments, illegal dumping, broken streetlights and dead animal removal.
The radar chart in Figure 22 details the average service request types in the resulting clusters
based upon the z-score.
69

Figure 20 Multivariate Clustering of 311 Data
70

Resulting charts from the analysis provided depth to the distribution of the MyLA311
data in regard to the clusters. The standardization of the frequency of the different service request
types discussed in Chapter 3 is necessary given the disparities within amounts reported, as
exemplified earlier in the chart from Figure 19. Visually and statistically identifying clear
differences in signatures of the clusters created from service requests are possible given the five
variables present in the multivariate clustering analysis. After the multivariate analysis of 311
service requests, Figure 20 displays four distinct groups with differing signatures. Resulting
tables and analysis of the data in Excel produced Figure 21 and Figure 22. Figure 21 details the
average service requests per cluster. Figure 22 serves the purpose of visually comparing the z-
scores of the different cluster signatures.

Figure 21 Average Service Requests per 311 Cluster
Cluster 1 had a high concentration of dead animal removals, followed by illegal dumping.
Cluster 2 had a high concentration of graffiti removal requests and homeless encampment
71

reports. Cluster 3 had no service request with a high concentration and was generally
characterized by a lack of reporting any type of service request. Finally, Cluster 4 had a high
concentration of reporting streetlight issues.

Figure 22 Radar chart of 311 service request cluster characteristics
The breakdown of the service requests is exemplified in Figure 23 and Table 7, both of
which are provided as results of the multivariate clustering analysis within ArcGIS Pro. Within
Table 7, the R-squared value for broken streetlights had the highest value at .625225, followed
by dead animal removal at .528803, graffiti removal at .51247, illegal dumping at .31403, and
finally homeless encampments at .241207. The R-squared value indicates the most effective
variable for establishing clusters. It reflects the amount of variation in the original data retained
after the clustering process; the larger the R-squared value is for a given variable, the better that
variable is at discriminating among selected features for indication of a cluster type.
72

Figure 23 Multivariate Clustering Box Plots – 311 Data

Table 7 Multivariate Clustering Table Results – 311 Data

4.2.1. 311 Clusters and Resulting Neighborhoods
Multivariate cluster analysis of 311 service requests in census tracts produced the
resulting clusters and signatures. Communities in Cluster 1 seemed to be most concentrated
within South Los Angeles and the Valley. Communities in Cluster 2 were generally located
around the Downtown, East Los Angeles area and Venice neighborhoods with some adjacent
clusters located in the Valley. It is worth noting that the dark colored census tracts in Figure 18
73

showed the areas submitting the most 311 service requests, and these same census tracts also
comprised Cluster 2. Problems in Cluster 2 communities are not simply reported, but heavily
reported. Out of all cluster profiles, Cluster 2 exemplified the most extreme frequency of reports
for graffiti removal and homeless encampments, with illegal dumping as the third highest
priority. Many of the communities in that fell in Cluster 2 were experiencing homelessness as a
wicked problem, heavily affecting the quality of life for residents according to service request
data.
Communities in Cluster 3 generally had the lowest reporting rates out of all cluster
profiles. While each service request type remained below the standard deviation for frequency of
reports, the frequency for reporting homeless encampments was the highest. Generally, these
areas either simply did not experience many problems, or less likely, did not report many of the
problems and dealt with them in other ways. Regardless of the reasoning behind the reporting
patterns of 311 data for the neighborhoods in Cluster 3, insight to how the built environment
functioned was still gained from observing the signature’s low service request intensity.
Finally, Cluster 4 was comprised of the fewest census tracts. This cluster profile was
unexpected given its very high frequency of reporting broken streetlights, something that
characterized no other cluster profile. There was no discernible pattern of the pockets of census
tracts belonging to Cluster 4, save for several census tracts grouped together in the Hollywood
Hills area. No census tracts fell into Cluster 4 down the Harbor Freeway and in the San Pedro
area.
4.3. Sociodemographic Attributes of Resulting 311 Clusters
The clusters created from the multivariate clustering of MyLA311 data provided insight
regarding the sociodemographic features of communities that submitted service requests. There
74

were some surprises found, mainly regarding the spatial distribution of 311 data clusters. Since
this study only looked at a portion of the possible 311 service requests from 2018, performing
this analysis with more types of service requests would provide additional insight to the
problems experienced by different communities in Los Angeles. Ultimately, the information
revealed by the sociodemographic clustering must be taken in conjunction with the information
revealed about a census tract’s 311 urban signature to provide the most detailed information
about a location at the local level.
4.3.1. Sociodemographic Cluster Characteristics
In order to define the urban signature of the resulting 311 clusters, the sociodemographic
data from census tracts must be taken into account. General insight was provided from the maps
procured for the study area in Chapter 1 from the same data. The multivariate clustering of
MyLA311 data proves beneficial in visualizing and determining the multiple sociodemographic
characteristics and service requests that make up an urban signature, as exemplified by the radar
chart in Figure 24 and corresponding z-score values in Table 8.
75

Figure 24 Radar Chart of Sociodemographic data from 311 Clustering
76

Table 8 Z-Scores of Cluster Characteristics

The breakdown of racial composition of resulting clusters seemingly mirrors the
clustering of groups identified by Wang et al. (2017) in the analysis. Cluster 1 had a high
disposition of Hispanic and Non-Hispanic African American residents, with barely any Non-
Hispanic white residents. Cluster 2 had mainly Non-Hispanic Asian residents, closely followed
by Hispanic residents. Cluster 3 had the most equal distribution of requests from residents of
different races and ethnicities. Cluster 4 was comprised of mainly Non-Hispanic white
communities, with rarely any Hispanic communities.
The sociodemographic information regarding the highest level of education attained
relates to the financial information, namely because of the correlation between education and
income dependent on employment. Cluster 1 had the highest amount of those with only a high
school education. While Cluster 1 also had the highest number of individuals with an associate
degree, it also had the lowest response rate regarding a bachelor’s degree or any graduate
education. Cluster 2 followed Cluster 1 with second highest attainment of only a high school
education, and had fewer associate degrees, yet more bachelor’s and graduate education. Cluster
3 was the second most overall educated group, falling only behind Cluster 4. Able to separate
itself from the other clusters, Cluster 4 had noticeably more four-year college and graduate
school levels of education than all other groups. Three trends were identifiable in regard to
financial situations. Cluster 1 and Cluster 2 generally had lower income, while Cluster 3 was
Cluster
Hispanic
NonHispanic White
NonHispanic African
American
NonHispanic Asian
Highschool
Associates
Bachelors
Graduate
Below Poverty
Med Income Nonfamily
Mean Income All
Med Inc HouseHold
Med Inc Fam
Unemployment Rate
Percent Uninsured
1 0.2513 -0.2099 0.2659 -0.3651 0.2885 0.0593 -0.3018 -0.2820 0.0178 -0.2493 -0.2002 -0.1548 -0.1776 0.0750 -0.0238
2 0.1176 -0.1237 -0.1807 0.2833 -0.0243 -0.1860 0.0753 -0.1345 0.0531 0.0033 -0.1656 -0.1636 -0.1685 -0.0777 0.3499
3 -0.1124 0.0992 -0.0999 0.1457 -0.0961 0.0394 0.1196 0.1291 -0.0087 0.1037 0.0881 0.0712 0.0793 -0.0093 -0.0288
4 -0.5607 0.6670 0.0397 -0.1562 -0.5852 -0.3288 0.4788 0.8232 -0.0629 0.5361 0.9435 0.8438 0.9019 -0.0454 -0.3766
77

regarded as the “middle class” of the four clusters. Communities belonging to Cluster 4 were
noticeably wealthier in all economic categories.
Percent uninsured, unemployment rate and below poverty were the remaining variables
without different types. Overall, there was no major difference in the percentage below poverty
and unemployment rate determined through the MyLA311 reporting clusters. The most
noticeable difference is visible in the percentage of uninsured individuals. Cluster 2 had the
highest rate of uninsured individuals, while Cluster 1 and Cluster 3 were generally around the
same percentage of uninsured respondents. Cluster 4 had the lowest percentage of individuals
without insurance; this is linked to the relation of income to insurance, discussed in Chapter 1.
78

Chapter 5 Conclusions and Discussion
This chapter concludes this thesis and provides a discussion about the objectives of the research
through a discussion of the results, their significance, limitations and future research. This
research analyzed the results of multivariate clustering of MyLA311 data for service requests
and the sociodemographic characteristics present in the resulting clusters. Resulting clusters’
sociodemographic variables were then analyzed through correlation statistics.
5.1. Significance of Findings
The analysis confirms that MyLA311, Los Angeles’s 311 service request data, can be
used to identify urban signatures within City of Los Angeles. Ultimately, this data reveals
geospatial patterns relating to citizens who made service requests and their built environment.
The findings and use of the MyLA311 data as VGI for geospatial analysis reveals new
information at the local scale. It is important to utilize the separate-and-combine approach for
understanding the features present in all of the data due to the variant patterns in subsets of the
census tracts (Minkoff 2015). This approach addresses the variations in patterns over each subset
and works best with the subsets’ respective analysis independent of each other.
This research also contributes to the field of analysis using VGI from city service
requests. The analysis provides another example where 311 service request data can be used
alongside sociodemographic data for a multivariate cluster analysis revealing spatial patterns at a
granular scale within an urban built environment specific to Los Angeles in comparison with the
analysis of Wang et al. (2017). This thesis provides an alternative methodology for the process
through Esri and Excel software instead of through R. The clusters formed appeared to be
successful in identifying neighborhood level clusters of differences in reporting frequency and of
sociodemographic characteristics.
79

5.1.1. Proof of Concept
The results of this study with MyLA311 data are meaningful for identification of urban
signatures at the local scale within a city. This analysis was not intended to produce a model of
311 service requests and the community profiles but to provide a proof of concept for using
MyLA311 service request data in analysis of the 311 service requests in describing urban activity
and characteristics, allowing the data to be taken into consideration for future models. A
stakeholder could examine the areas of interest in the resulting maps to gain a better
understanding of the human context alongside concerns presented through 311 data. The analysis
benefits using MyLA311 data is the capability of identifying the correlations between service
requests and sociodemographic characteristics at a finer scale.
5.1.2. Implications for Los Angeles
The results of this analysis provide stakeholders with clearer pictures about the
constituents and needs of an area. For example, a city council district can observe the different
clusters in neighborhoods and the differences present within those boundaries. Given the
granular level of a census tract, the patterns observed could show a concentrated image of
community makeup and offer insight as to which problems are being reported the most by
citizens. In theory, a council district’s local governing body can then tackle those problems at the
source and make a concentrated effort to alleviate them at the root cause. While this research
solely focused on MyLA311 data collected in 2018, the data is available from 2016 to the
present. This larger temporal scope of data can be analyzed to detect changes in the needs of a
community and its constituents over time.
80

5.2. Study Limitations and Future Research
While this study successfully used MyLA311 data to identify urban signatures within
City of Los Angeles, there were several observed limitations in this study. This section addresses
these limitations and what work can be done in the future to improve upon these results.
5.2.1. Limitations
In creating a methodology for the multivariate cluster analysis in ArcGIS, there were
minor problems relating to sliver polygons and ensuring only relevant information is taken into
account. The main limitation of this study relates to the source of the data in how MyLA311 data
includes only reported service requests, not every occurrence in the communities within City of
Los Angeles. Thus, the multivariate clustering results were only based upon reported problems.
As a result, the frequencies may not be 100% representative for the area. As a result, it must be
accepted that the data serve as a proxy for the actual occurrences of service requests in the area.
While this does not detract from the success of the methodology towards defining an urban
signature, it can influence the clustering results.
An additional caveat stemming from the 311 service requests is the reporting patterns of
individuals. Some individuals may be considered power users of reporting service requests and
report at a high rate, whereas other individuals may rarely report or not at all. Kontokosta (2017)
claims that disparity in reporting stems from a variety of problems, but mainly a lack of interest
of interaction with the government, varying levels of trust regarding government agencies, or a
lack of accessibility through technology
Moreover, only few types of service requests were available through the LA Open Data
Portal, while many exist within the restricted backend of MyLA311’s server. This extends to the
politics of only specific MyLA311 service request type data’s availability through the Open Data
81

Portal. While the city may tout its commitment to transparency and publicly available data, there
is a need for more demonstrations of this commitment. In relation to this analysis, incorporating
more service request types would assist in creating signature clusters based off of additional,
physically prominent service request types.
Concurrently, the bias of citizen reporting can prejudice the frequency of specific reports,
as the data for reports are based on what an individual has a propensity to report: for example,
Individual A may report all problems that they see, whereas Individual B has a penchant for only
reporting homeless encampments. While VGI through civic service requests is an asset for
transforming civic activism into big data, it often begs the question regarding quality of data. The
publicly available MyLA311 data does not include if a resident who lives near the service
request location submitted it, or if it was submitted by someone traveling through the area,
perhaps for work or leisure, who felt compelled enough to submit a service report. While it is
impossible to arrive at a conclusion for who exactly is submitting the service request, the
locational component of the service request provides insight into where people feel the need to
fix physical problems.
At the time of publishing, the full list of 311 service request types is not available on
MyLA311. As the city undergoes more advancements in technological capabilities and data
management, it is expected that more data will be made available to the public. This in turn will
lead to more opportunities for inclusion within a spatial analysis using City of Los Angeles
proprietary data.
5.2.2. Future Research
In the future, it would be beneficial for the same analysis to be done with additional
service requests in order to provide a more detailed picture of what problems citizens are
82

experiencing, as well as compare clustering from service requests in previous years. Given the
vast options for service requests, if all types of service requests were available to the public on
the Open Data LA website, it would then make sense to make smaller groups for multivariate
clustering of service requests that are thematically similar, such as only refuse or investigation
service requests listed in Table 1, so as not to provide an overload of information ultimately not
revealing granular details in the cluster results. It would be also beneficial to analyze multiple
years of 311 data to observe the changes over time within the census tracts.
This thesis serves as a baseline to provide a methodology to observe and analyze clusters
of MyLA311 service requests. Future research direction can follow the existing research from
other cities that uses the same type of 311 data to perform predictive analytics within a city. Lu
(2016) utilizes a comparison of relative request share for each channel, a spatial hot spot
analysis, and a regression model to compare 311 service type usages alongside
sociodemographic variables. Performing these types of analysis with the 311 data for Los
Angeles would prove beneficial to multiple stakeholders and offer additional proof for the
modernization of data-driven city services and technology. Zha (2014) provides a methodology
for using current 311 service request patterns for a prediction of future 311 service requests,
providing a city with insight regarding budgeting of resources for future services. Should this
analysis be performed with the MyLA311 Los Angeles data, it could provide a basis for
preemptive problem solving in the city to saving time, money and other resources.
5.3. Conclusion
In conclusion, multivariate clustering of Los Angeles MyLA311 service requests
produces clusters reminiscent of an urban signature to provide more insight as to the needs and
problems facing certain communities in Los Angeles. When coupled with the sociodemographic
83

attributes in the census data, such service request clusters provided a more detailed
characterization of the local communities. Identifying which census tracts or neighborhoods
experience a low total amount of service requests could be indicative of a need to advertise the
services more to the community, or indicative of less existing problems available to report. This
sort of spatial analysis provides new context and understanding of an area and how the
community interacts with the built environment based on observed knowledge of the 311 service
requests at the local scale and the correlating neighborhood characteristics.
The analysis provides quantitative understanding of the built environment and makeup of
the city through data relating to the communities inhabiting a space. By utilizing the citizen
produced data of MyLA311, urban planners and policy-makers have the opportunity to approach
civic issues with renewed insight and data to support any proposals or plans. Replication of this
research within Los Angeles should address all available service requests rather than the five
determined in this analysis as a proof of concept. Finally, by providing more insight to the
makeup of communities and better understanding the problems residents face, Los Angeles
policy makers and local government institutions can utilize data-driven analytics to best address
problems while taking the human component into consideration.
84

References
Allen, James P. and Eugene Turner. 2002. Changing Faces, Changing Places: Mapping
Southern Californians. California State University, Northridge. Center for Geographical
Studies
Antoniou, V., & Skopeliti, A. 2015. “Measures and Indicators of VGI Quality: An
Overview.” ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial
Information Sciences 2, 345.
Bandura, Albert. 2000. “Exercise of Human Agency Through Collective Efficacy.” Current
Directions in Psychological Science 9(3), 75-78.
Bégin, D., Devillers, R., & Roche, S. 2013. “Assessing Volunteered Geographic Information
(VGI) Quality Based on Contributors’ Mapping Behaviours.” Paper presented at
the Proceedings of the 8th International Symposium on Spatial Data Quality ISSDQ, 149-
154.
Bengfort, Jacquelyn. 2019. "Cities Move 311 Systems to the Cloud and Improve Citizen
Services." State Tech Magazine., last modified January 11, accessed January 15, 2019,
https://statetechmagazine.com/article/2019/01/cities-move-311-systems-cloud-and-
improve-citizen-services.
Bettencourt, Luís MA, José Lobo, Deborah Strumsky, and Geoffrey B. West. 2010. "Urban
Scaling and its Deviations: Revealing the Structure of Wealth, Innovation and Crime
Across Cities." PloS One 5 (11): e13541.
Brovelli, Maria Antonia, Marco Minghini, and Giorgio Zamboni. 2015. “Public Participation
GIS: A FOSS Architecture Enabling Field-data Collection.” International Journal of
Digital Earth 8(5), 345-363.
Brown, Gregory, Maggi Kelly, and Debra Whitall. 2014. “Which ‘public'? Sampling Effects in
Public Participation GIS (PPGIS) and Volunteered Geographic Information (VGI)
Systems for Public Lands Management.” Journal of Environmental Planning and
Management 57(2), 190-214.
Bundorf, M. Kate and Mark V. Pauly. 2006. "Is Health Insurance Affordable for the Uninsured?"
Journal of Health Economics 25 (4): 650-673.
Chamard, Sharon. 2010. Homeless Encampments. US Department of Justice, Office of
Community Oriented Policing Services Washington, DC.
Collins, Brady and Anastasia Loukaitou-Sideris. 2016. "Skid Row, Gallery Row and the Space in
between: Cultural Revitalisation and its Impacts on Two Los Angeles Neighbourhoods."
Town Planning Review 87 (4): 401-427.
Composto, Sarah, Jens Ingensand, Marion Nappez, Olivier Ertz, Daniel Rappo, Rémi Bovard,
Ivo Widmer, and Stéphane Joost. 2016. “How to Recruit and Motivate Users to Utilize
85

VGI-systems.” Paper, 19th AGILE Conference on Geographic Information Science,
Helsinki, Finland.
Currie, Morgan. 2016. “The Data-Fication of Openness the Practices and Policies of Open
Government Data in Los Angeles.” PhD diss., UCLA.
Dear, Michael. 1992. "Understanding and Overcoming the NIMBY Syndrome." Journal of the
American Planning Association 58 (3): 288-300.
Early, Dirk W. 2005. “An Empirical Investigation of the Determinants of Street Homelessness.”
Journal of Housing Economics 14(1), 27-47.
Esri. 2016. “Los Angeles Launched GeoHub.” Accessed February 2, 2018. Retrieved from
https://www.esri.com/esri-news/arcnews/spring16articles/los-angeles-launched-geohub
Esri. 2018. “How Multivariate Clustering works.” Accessed December 21, 2018. Retrieved from
https://pro.arcgis.com/en/pro-app/tool-reference/spatial-statistics/how-multivariate-
clustering-works.htm
Ester, Martin, Hans-Peter Kriegel, Jörg Sander, and Xiaowei Xu. 1996. “A Density-Based
Algorithm for Discovering Clusters in Large Spatial Databases with Noise.” Kdd 96 (34),
226-231.
Federal Communications Commission. 1997. Use of N11 Codes and Other Abbreviated Dialing
Arrangements. Washington, DC. Accessed on March 14,2019.
https://transition.fcc.gov/Bureaus/Common_Carrier/Orders/1997/fcc97051.pdf
Flanagin, Andrew J. and Miriam J Metzger. 2008. “The Credibility of Volunteered Geographic
Information.” GeoJournal 72(3-4), 137-148.
Fonte, C. C., Lucy Bastin, G. Foody, T. Kellenberger, N. Kerle, P. Mooney, A-M Olteanu-
Raimond, and L. See. 2015. “VGI Quality Control.” ISPRS Annals of Photogrammetry,
Remote Sensing and Spatial Information Sciences, 2, 317-324.
Gehl, J. 2010. Cities for People. Washington, DC: Island Press.
Goodchild, Michael F. 2007. “Citizens as Sensors: The World of Volunteered Geography.”
GeoJournal 69(4), 211-221.
Goodchild, Michael F. and Linna Li. 2012. “Assuring the Quality of Volunteered Geographic
Information.” Spatial Statistics 1, 110-120.
Handy, Susan L., Marlon G. Boarnet, Reid Ewing, and Richard E. Killingsworth. 2002. “How
the Built Environment Affects Physical Activity: Views from Urban Planning.” American
Journal of Preventive Medicine, 23(2), 64-73.
Hood, Christopher. 1995. “The ‘new public management’ in the 1980s: Variations on a theme.”
Accounting, Organizations and Society, 20(2-3), 93-109.
86

Johnson, Steven. 2010. "What a Hundred Million Calls to 311 Reveal About New York." Wired
Magazine.
Koenig, John F. 2001. "Spaces of Denial and Denial of Place: The Architectural Geography of
Homelessness in Victoria, BC." Master's thesis, University of Victoria.
Kontokosta, Constantine E. and Awais Malik. 2018. “The Resilience to Emergencies and
Disasters Index: Applying Big Data to Benchmark and Validate Neighborhood Resilience
Capacity.” Sustainable Cities and Society, 36, 272-285.
Kontokosta, Constantine, Boyeong Hong, and Kristi Korsberg. 2017. “Equity in 311 Reporting:
Understanding Socio-Spatial Differentials in the Propensity to Complain.” Presented at
the Bloomberg Data For Good Exchange.
Los Angeles City Controller. 2019. Los Angeles City Controller Control Panel. Accessed
December 12, 2018. https://controllerdata.lacity.org/
Los Angeles County Department of Public Health. 2017. Recent Trends in Health Insurance
Coverage in Los Angeles County. Office of Health Assessment and Epidemiology.
http://publichealth.lacounty.gov/docs/LaHealth_RecentTrendsInHealthInsuranceCoverag
e_yr2017.pdf
Lu, Qing and Peter A. Johnson. 2016. “Characterizing New Channels of Communication: A
Case Study of Municipal 311 Requests in Edmonton, Canada.” Urban Planning, 1(2), 18-
31.
Lulka, David. 2013. “The Posthuman City: San Diego's Dead Animal Removal Program.” Urban
Geography 34(8), 1119-1143.
Matsumoto, Shigeru and Kenji Takeuchi. 2011. “The Effect of Community Characteristics on
the Frequency of Illegal Dumping.” Environmental Economics and Policy Studies 13(3),
177-193.
McCormack, Gavin R., Melanie Rock, Ann M. Toohey, and Danica Hignell. 2010.
“Characteristics of Urban Parks Associated with Park use and Physical Activity: A
Review of Qualitative Research.” Health & Place 16(4), 712-726.
Megler, Veronika, David Banis, and Heejun Chang. 2014. “Spatial Analysis of Graffiti in San
Francisco.” Applied Geography 54, 63-73.
Minkoff, Scott L. 2016. “NYC 311: A Tract-Level Analysis of Citizen–Government Contacting
in New York City.” Urban Affairs Review 52(2), 211-246.
Moon, M. Jae. 2002. "The Evolution of E-government among Municipalities: Rhetoric Or
Reality?" Public Administration Review 62 (4): 424-433.
87

Morenoff, Jeffrey D., Robert J. Sampson, and Stephen W. Raudenbush. 2001. “Neighborhood
Inequality, Collective Efficacy, and the Spatial Dynamics of Urban
Violence.” Criminology 39(3), 517-558.
Mullen, William F., Steven P. Jackson, Arie Croitoru, Andrew Crooks, Anthony Stefanidis, and
Peggy Agouris. 2015. “Assessing the Impact of Demographic Characteristics on Spatial
Error in Volunteered Geographic Information Features.” GeoJournal, 80 (4), 587–605.
Nam, Taewoo. 2012. “Modeling Municipal Service Integration: A Comparative Case Study of
New York and Philadelphia 311 Systems.” Dissertation, University at Albany, State
University of New York.
Nam, Taewoo and Theresa Pardo. 2014. “The Changing Face of a City Government: A Case
Study of Philly311.” Government Information Quarterly 31. 10.1016/j.giq.2014.01.002.
Neis, Pascal, Dennis Zielstra, and Alexander Zipf. 2012. “The Street Network Evolution of
Crowdsourced Maps: OpenStreetMap in Germany 2007–2011.” Future Internet 4: 1– 21.
Novak, Kurt. 1995. “Mobile Mapping Technology for GIS Data Collection.” Photogrammetric
Engineering and Remote Sensing 61 (5): 493-501.
O’Brien, Daniel Tumminelli, Dietmar Offenhuber, Jessica Baldwin-Philippi, Melissa Sands, and
Eric Gordon. 2017. “Uncharted Territoriality in Coproduction: The Motivations for 311
Reporting.” Journal of Public Administration Research and Theory 27(2), 320-335.
O'Brien, Daniel Tumminelli. 2016. “Using Small Data to Interpret Big Data: 311 Reports as
Individual Contributions to Informal Social Control in Urban Neighborhoods.” Social
Science Research 59: 83-96.
Office of Los Angeles Mayor Eric Garcetti. 2013. Executive Directive No 3. Retrieved January
30, 2019, from https://www.lamayor.org/sites/g/files/wph446/f/page/file/Executive-
Directive-3-Open-Data.pdf?1426620075
Pain, Rachel, Robert MacFarlane, Keith Turner, and Sally Gill. 2006. “‘When, Where, if, and
but’: Qualifying GIS and the Effect of Streetlighting on Crime and Fear.” Environment
and Planning A 38(11), 2055-2074.
Sampson, Robert J., Stephen W. Raudenbush, and Felton Earls. 1997. “Neighborhoods and
Violent Crime: A Multilevel Study of Collective Efficacy.” Science 277(5328), 918-924.
Sayad, Saed. 2019. “K-Means Clustering.” Data Mining Map. Accessed March 18.
https://www.saedsayad.com/data_mining_map.htm.
Schwester, Richard W., Tony Carrizales, and Marc Holzer. 2009. "An Examination of the
Municipal 311 System." International Journal of Organization Theory & Behavior 12
(2): 218-236.
Scruggs, Carl. 2013. Best practices in open data initiatives. Office of Legislative Oversight.
88

Sieber, Renee E., and Peter A. Johnson. 2015. “Civic Open Data at a Crossroads: Dominant
Models and Current Challenges.” Government Information Quarterly 32(3), 308-315.
Tauberer, Joshua. 2014. Open Government Data: The Book. 2
nd
ed.
Tobler, Waldo R. 1970. "A Computer Movie Simulating Urban Growth in the Detroit Region."
Economic Geography 46, no. sup1: 234-240.
Toregas, Costis. 2001. “The Politics of E-Gov: The Upcoming Struggle for Redefining Civic
Engagement.” National Civic Review 90 (3): 235-240.
Trevino, Andrea. 2016. “Introduction to K-Means Clustering.” Oracle DataScience. Retrieved
March 30, 2019, from https://www.datascience.com/blog/k-means-clustering.
Trinacria, Joe. 2018. “Pew Report: Philly Remains the Poorest of America's 10 Largest Cities.”
Philadelphia Magazine. April 4. https://www.phillymag.com/news/2018/04/06/pew-
report-poverty/.
U.S. Census Bureau 2018. American Community Survey, 2017 American Community Survey 5-
Year Estimates using American FactFinder. http://factfinder2.census.gov
Wang, Lingjing, Cheng Qian, Philipp Kats, Constantine Kontokosta, and Stanislav Sobolevsky.
2017. “Structure of 311 Service Requests as a Signature of Urban Location.” PloS
One 12(10): e0186314.
Wessel, Nate. 2016. “Assessing Predictors of Citizen Reports of Dead Animals in Cincinnati.”
Paper, University of Toronto.
Whitaker, Gordon P. 1980. “Coproduction: Citizen Participation in Service Delivery.” Public
Administration Review 240-246.
White, Ariel and Kris-Stella Trump. 2018. “The Promises and Pitfalls of 311 Data.” Urban
Affairs Review 54(4), 794-823.
Williams, C. H. 2007. “The Built Environment and Physical Activity: What is the Relationship?”
Report, University of Indianapolis.
Wilson, James Q. and George L. Kelling. 1982. “Broken windows.” Atlantic Monthly 249(3), 29-
38.
Zha, Yilong Frank and Manuela Veloso. 2014. “Profiling and Prediction of Non-Emergency
Calls in NYC.” Semantic Cities: Beyond Open Data to Models, Standards and
Reasoning: In Workshops at the AAAI-14 Conference on Artificial Intelligence.

Abstract (if available)

Abstract In order to increase citizen engagement, in 2013, the City of Los Angeles introduced the MyLA311 application, a smartphone app that allows residents to easily request city services. Previous service requests were funneled through four separate data service management systems and lacked transparency

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Evaluating the relationship between Colorado elk hunting success and terrain ruggedness

PDF

Redefining urban food systems to identify optimal rooftop community garden locations: a site suitability analysis in Seattle, Washington

PDF

Urban green space accessibility and environmental justice: a GIS-based analysis in the city of Phoenix, Arizona

PDF

Soil lead contamination from the Exide battery smelter: the role of spatial scale in cleanup efforts

PDF

Does the Bay Area have a social center? Delimiting the postmodern urban center of the San Francisco Bay Area

PDF

Collecting and managing VGI infrastructure assessments in support of stability operations

PDF

The role of amenities in measuring park accessibility: a case study of Downey, California

PDF

Developing a replicable approach for the creation of urban climatic maps for urban heat island analysis: a case study for the city of Los Angeles, California

PDF

A methodology for a real estate blockchain application utilizing geographic information systems (GIS)

PDF

A spatial investigation of New York City's historical shoreline

PDF

A comparison of urban land cover change: a study of Pasadena and Inglewood, California, 1992‐2011

PDF

Spatial analysis of vision services of Kaiser Permanente members

PDF

California ballot results viewer, 2008-2018: a Web GIS application for viewing ballot proposition results in California

PDF

An exploratory spatial analysis of fire service and EMS accessibility in northeastern Illinois communities

PDF

Finding the green in greenspace: an examination of geospatial measures of greenspace for use in exposure studies

PDF

A geospatial analysis of income level, food deserts and urban agriculture hot spots

PDF

Urban areas and avian diversity: using citizen collected data to explore green spaces

PDF

Distribution of Sonoran pronghorn (Antilocapra americana sonoriensis) on an active Air Force tactical range

PDF

Philly Bike Report: a mobile app for mapping and sharing real-time reports of illegally blocked bike lanes in Philadelphia

PDF

Practical application of ACS place of birth data in an app created for American Red Cross International Services

Asset Metadata

Creator Windisch, Richard Andrew (author)

Core Title Utilizing 311 service requests as a signature of urban location in the City of Los Angeles

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 07/24/2019

Defense Date 05/21/2019

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

Tag GIS,multivariate clustering,OAI-PMH Harvest,service request,urban signature

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Sedano, Elisabeth (committee chair), Lee, Su Jin (committee member), Wu, An-Min (committee member)

Creator Email richwindisch@gmail.com,rwindisc@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-189102

Unique identifier UC11663162

Identifier etd-WindischRi-7604.pdf (filename),usctheses-c89-189102 (legacy record id)

Legacy Identifier etd-WindischRi-7604.pdf

Dmrecord 189102

Document Type Thesis

Format application/pdf (imt)

Rights Windisch, Richard Andrew

Type texts

Source 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 a...

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