University of Southern California Dissertations and Theses

Spatiotemporal hotspots of 2018-2020 crime in Houston, Texas

(USC Thesis Other)

Spatiotemporal hotspots of 2018-2020 crime in Houston, Texas

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Spatiotemporal Hotspots of 2018-2020 Crime in Houston, TX

by

Geoffrey J. Shreve

A Thesis Presented to the
FACULTY OF THE USC DORNSIFE COLLEGE OF LETTERS, ARTS AND SCIENCES
University of Southern California
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOGRAPHIC INFORMATION SCIENCE AND TECHNOLOGY)

December 2022

Copyright © 2022 Geoff Shreve

ii

Acknowledgements

I would like to thank my parents, relatives and my brother for their support. I would also like to
thank my committee chair, Dr. Bernstein for all her help and guidance throughout the thesis
development process. Lastly, I would like to thank my committee members, Drs. Wilson and
Ruddell, for their input and guidance especially in the early stages of the thesis development and
towards the end when preparing, defending and revising the final draft.

iii

Table of Contents

Acknowledgements……………………………………………………………………………..…ii
List of Tables…………………………………………………….……………………………….vi
List of Figures…………...…………………………………………………………………….....vii
Abbreviations……………………………………………………………………………………viii
Abstract………………………………………………………………………………………..….ix
Chapter 1 Introduction………………………………………………………………………….…1
1.1 Project Scale and Study Area……………………..…………………………………..…...1
1.2 Data………………………………………………………..…………………………..…..2
1.3 Motivation……………………………………………………………………………...….5
1.4 Research Goals……………………………………..……………………………...............6
1.5 Thesis Structure………………………...............................................................................7
Chapter 2 Related Work…………………………………………………………………………...8
2.1 Temporal and Spatial Extents……………………………………………………..…....…8
2.2 Place and Crime……………………………………….......................................................9
2.3 The Use of Innovative Methods in Crime Analysis………………………………….......10
Chapter 3 Research Design and Methods………………………………………………………..13
3.1 Description of Crime Attributes…………………………………..……………………...14
3.2 Organization of the Crime Data…………………………..……………………………...15
3.3 Geocoding of Crime Addresses ……………………………………………………........16
3.3.1 Geocoding in Excel using CDX WinZip………………………………………..…16
3.3.2 Geocoding in Google Sheets using Geocoding by Smart Monkey……………..….17
3.3.3 Geocoding in ArcGIS Pro……………………………………………………...…..18
iv

3.4 Spatial Analysis of the Crime Data……………………………………………..………..18
3.5 Land Use and Optimized Hotspots……………………………………………..…..……20
Chapter 4 Results………………………………………………………………………………...22
4.1 Crime Trends and Results of all Crime Type hotspots…………………………………..22
4.1.1 2018-2020 Assault Hotspots…………………………………………………….…22
4.1.2 2018-2020 Summary of Burglary Hotspots………………………………………..30
4.1.3 2018-2020 Summary of Robbery Hotspots………………………………………..34
4.1.4 2018-2020 Summary of Theft Hotspots…………...................................................40
4.1.5 Crime Types by Year………………………………………………………………45
4.2 2020 Land Use Types for Optimized Crime Hotpots……….….....……….………….....45
4.3 Crime Statistics Before and After Natural Disasters…………….…................................52
4.4 Bus Hotspot Locations and 2020 Crime Hotspot Locations……………………….....….53
4.5 2020 Edge Effects Test of Data: July-December 2020………………………………..…54
4.5.1 2020 Assault Boundary Hot Spots…………………………………………………54
4.5.2 2020 Burglary Boundary Hot Spots…………………………………………….….55
4.5.3 2020 Robbery Boundary Hot Spots………………………………………………..56
4.5.4 2020 Theft Boundary Hot Spots…………………………………………………...57
Chapter 5 Conclusions…………………………………………………………………………...59
5.1 Research Questions and Summary of Findings ………………………......………..……60
5.1.1 Crime hot spots located within the city …………………………………………....60
5.1.2 What types of crime constitute different hot spots?.................................................61
5.1.3 Crime trends for 2018-2020……………………………………………...……......61
5.1.3.1 Assault Crime Hotspots………….........................................................................61
v

5.1.3.2 Burglary Crime Hotspots……………………………………………………..….62
5.1.3.3 Robbery Crime Hotspots……………………………………………………...….62
5.1.3.4 Theft Crime Hotspots……………………………………………………….……63
5.1.4 2020 Land Use Types and Optimized Crime Hotpots………………………..…....64
5.1.5 Crime and Natural Disasters……….……………………………………...…..…...66
5.2 Successes, Challenges and Failures………………………………………………..…….67
5.3 Implications for Law Enforcement…………………………………………………..…..69
5.4 Future Research………..…………………………..…………………………………….70
References…………………………………………………………………………………...…...74
Appendices……………………………………………………………………………………….76
Appendix A Bus Stop Density and 2020 Theft Hotspots………………………………....…76
Appendix B Bar Locations within the Study Districts………………………………..……...77
Appendix C Metro Bus Stop Density……………………………………………..…………78

vi

List of Tables

Table 1 Thesis Project Workflow………………………………………………………………..14
Table 2 2018-2020 Crime Statistics…………….……………..………………………………....22
Table 3 Crime Statistics Before and After Tropical Storm Beta……………………………..….52
Table 4 Crime Statistics Before and After Tropical Storm Imelda…………………………...…53
Table 5 Crime Statistics Before and After Hurricane Harvey…………………………………...53

vii

List of Figures

Figure 1 Study Police Jurisdictions………………………….………………………………….....3
Figure 2 Houston, TX Major Roads and Districts…………………………………………...........4
Figure 3 2018-2020 Assault Crime Hotspots.................................................................................24
Figure 4 Frequency of Assault Hotspots ………………………………………...……………....28
Figure 5 2018-2020 Burglary Crime Hotspots………………………………………………......31
Figure 6 Frequency of Burglary Crime Hotspots………………………………………………..33
Figure 7 2018-2020 Robbery Crime Hotspots …………..………………………………….…...35
Figure 8 Frequency of Robbery Crime Hotspots……………………………………………...…37
Figure 9 2018-2020 Theft Crime Hotspots.……………………………………….……..............41
Figure 10 Frequency of Theft Crime Hotspots………………………………………...…..….....44
Figure 11 2018 Crimes by Type……………………….……...…………….………………...…46
Figure 12 2019 Crimes by Type………………………………………….............................…...47
Figure 13 2020 Crimes by Type…………………………………………............................…....48
Figure 14 2020 Optimized Crime Hotspots……………………………………………….…......50
Figure 15 Downtown and Midtown District Land Uses……………………………………........51
Figure 16 July-December 2020 Edge Effects ……………………………...………………........58
Appendix A Bus Stop Density and 2020 Theft Hotspots……………………………………..…76
Appendix B Bar Locations within the Study Districts……..………….…………………………77
Appendix C Metro Bus Stop Density……………………………………………………………78

viii

Abbreviations

COHGIS City of Houston Geographic Information System
CSV Comma-separated value
FIPS Federal Information Processing Standards
GIS Geographic Information System
HPD Houston Police Department
LAPD Los Angeles Police Department
NIBRS National-Incident Based Reporting System
STAC Spatial and Temporal Analysis of Crime

NT-STAC Network Spatial and Temporal Analysis of Crime

NT-SaTScan Network Spatial and Temporal Scan

ST-DBSCAN Spatial Temporal Density Based Scan

STKDE Spatial Temporal Kernel Density Estimation
STSNN Spatial Temporal Shared Nearest Neighbors
STSSS Space-Time Scan Statistics
STW Spatio-temporal Window
WNN Windowed Nearest Neighbor

ix

Abstract

In 2018, Houston’s crime rate was higher than the rates in 95% of U.S. cities. Houston’s
population is the fourth highest in the nation with more than 2 million people, all of whom are
affected by this high crime rate. A better understanding of the spatial and temporal aspects of
crime would be useful for law enforcement in protecting the general population. This study
analyzed assaults, burglaries, robberies, and thefts in the inner Interstate 610 area of Houston,
which is considered downtown. The Houston police department provided crime address data for
each crime type from 2018 to 2020. The crime data was geocoded in ArcGIS Pro into point
shapefiles and aggregated using counts. The Esri Optimized Hot Spot Analysis and Kernel
Density Tools were used to determine crime hot spots for each crime type. The study also
explored whether land use type was related to hotspots of certain crimes in the Downtown and
Midtown districts of Houston. The study found that the crime hotspots for each crime type
occurred mainly in the Downtown, Midtown, and Montrose districts of Houston. Thefts and
assaults were higher near the downtown bar district. Theft was also higher near bus stations. The
study results could be valuable in helping the police predict and respond to crime hot spots in the
future in the Houston area, and the methods used may help the police manage crime in other
geographic areas and over different time periods as well.
1

Chapter 1 Introduction

Crime is a universal problem. Government agencies, and police departments in particular, want
to better understand crime rates in their jurisdictions: where it occurs, what occurs, and how
these attributes change over time. Geographic Information System (GIS) tools can be used to
better understand crime patterns and trends. At the most basic level, by using crime addresses
provided by police, these crime occurrences can be mapped spatially. Through using the kernel
density and other analysis tools, areas where more and less crime occurs and/or is likely to occur
can be identified. From there, one can look at whether or not these locations mirror other
features, such as bus stops, entertainment or industrial areas. This information can help police
departments trying to reduce crime to better understand past trends and to plan for the future.
This study identifies and analyzes crime hotspots over time in Houston, TX. The study
looks at four types of crime: assault, burglary, robbery and theft. A number of maps of different
crime types from 2018-2020 illustrate crime trends in the city over time. This study also explored
whether certain land uses and point features (e.g., bus stations, bars, etc.) are related with crime
types. All of these forms of analysis and visualization help a reader better understand Houston’s
crime rates and locations. This project helps to better explain crime type locations and changes
over time, thus enabling better crime protection in the region. Other police districts may want to
replicate this methodology should they have similar goals.
1.1 Project Scale and Study Area
The study area was Houston, TX. More specifically, it was the inner region of Interstate
610, which makes a loop around the central Houston metropolitan area. A small area outside
Interstate 610 was also analyzed for crime hotspots because crime still occurs on the other side of
the interstate. Houston was chosen as the study site because Houston’s crime rate in 2018 was
2

higher than the rates in 95% of U.S. cities (City-Data.com 2019). Thus, given this high crime
rate, this area seemed in particular need of attention. The temporal scale was 2018-2020, which
was chosen because it would provide the most recent data for law enforcement officials seeking
to use the results of this study to address crime in the city. While this is a very short window, it
can at least be suggestive of recent trends.
1.2 Data
A number of data sources were used in this study, which will be discussed in more detail
in Chapter 3. Most critical to this analysis was the crime location data. The Houston Police
Department (HPD) provided crime incidence data. There are 48 police jurisdictions within the
study area, and the crime points were organized and downloaded by police jurisdiction. This
included offense type, incident number, street number and name, and time and date for each
crime. The types of crime were assault, burglary, robbery, and theft. The crime occurrence data
was provided in an excel shapefile with street name and address. This research project could not
have been conducted without crime data at this level of precision.
Figure 1 shows a map of the Houston Police Department Police jurisdictions. I-610 is
identified to display the boundary between the inner police jurisdictions and the outer police
jurisdictions. The small inset map included in Figure 1 references the study area in relation to
Harris County. The area in yellow represents no data. This area is monitored by the Bel Air
Police Department, not the Houston Police Department. The study area was divided into four
district locations: the downtown, midtown, Montrose and Upper Kirby districts. These study
districts are shown in Figure 2.

F i g u r e 1 . M a p o f p o l i c e j u r i s d i c t i o n s w i t h i n t h e s t u d y a r e a . T h e a r e a i n y e l l o w h a s n o c r i m e
d a t a a n d t h e i n s e t m a p s h o w s t h e l o c a t i o n o f t h e s t u d y a r e a w i t h i n H a r r i s C o u n t y , T X .
1 i n c h e q u a l s 6 0 m i l e s
±
610
610
610
610
610
610
1A 10
1A 20
1A 30
1A 40
1A 50
2A 10
2A 20
2A 30
2A 40
2A 50
2A 60
3B 10
3B 30
3B 40
3B 50
5F10
7C 10
7C 20
7C 30
8C 10
8C 20
8C 30
9C 10
9C 20
9C 30
10H 10
10H 20
10H 30
10H 40
10H 50
10H 60
10H 70
10H 80
11H 10
11H 20
11H 30
11H 40
13D 10
13D 20
14D 10
14D 20
14D 30
15E 10
15E 30
15E 40
17E 10
18F10
18F20
0 2 .5 5 1 .2 5 M ile s
3
0 2 .5 5 1 .2 5 M ile s
4
6 1 0
F i g u r e 2 . M a p s h o w i n g t h e m a j o r r o a d s a n d t h e D o w n t o w n ( r e d ) , M i d t o w n ( o r a n g e ) , M o n t r o s e ( g r e e n ) ,
a n d U p p e r K i r b y D i s t r i c t s ( y e l l o w ) u s e d i n t h i s s t u d y . T h e i n s e t m a p s h o w s t h e l o c a t i o n o f t h e s t u d y a r e a
w i t h i n H a r r i s C o u n t y , T X .
1 i n c h e q u a l s 6 0 m i l e s
±
6 1 0
6 1 0
6 1 0
6 1 0
6 1 0
W a s h in g t o n A v e n u e
4 5
2 8 8
9 0
9 0
6 9
6 9
4 5
1 0
1 0
5

The city of Houston GIS (COHGIS) also provided a land use layer for the study, which
included ten land use categories. These data were used to look at the relationship between four
specific land use types (single-family, multi-family, commercial, and undeveloped) and high or
low crime rates. COHGIS also provided a bus stop layer recording the locations of bus stops
within the study area. Past research has shown a relationship between bus stops and crime, and
thus this data allowed the relationship to be statistically analyzed.
1.3 Motivation
A spatial understanding of crime hotspots can help prevent loss of life, loss of property
and material possessions, and damage to resources generally. The creation of this hotspot
analysis for Houston, TX was motivated by the desire to equip law enforcement agencies and
organizations looking to deter, suppress, and prevent crime with the tools they need to do so. If
the police are aware of the highest crime areas, then they can position crews and patrols in
predetermined areas susceptible to higher rates of crime. By showing the specific crime type
prevalent in particular areas, which this thesis does, police can better anticipate specific crime
types in specific areas. This type of approach can be applied to other metropolitan areas beyond
Houston should an analyst have access to similar data.
The second motivation is to inform the general public, especially those residing or
visiting the Houston area of crime patterns. If the public is aware of crime hotspots, they may be
less likely to enter high crime areas and more likely to take precautions, potentially alleviating
pressure on law enforcement. This study could also allow the public to become more vigilant and
watchful of suspicious activities in their communities. A reduction in crime can benefit a
community by increasing property values, reducing the time and economic cost of crimes that go
to trial, and ultimately, increasing the safety of individuals in the area.
6

1.4 Research Goals
This study had five primary research goals. The first goal was to determine where crime
hot spots are located within the city of Houston. The geocoded crime points were used to
determine the crime hot spots. The crime hot spots were divided into five classes: very high,
high, medium, low, and non-existent. The non-existent hotspots were not included in the study
because they were not areas of interest to the target audience, law enforcement.
The second goal of the study was to determine what type of crime occurred in different
hot spots. The crime types were analyzed and mapped separately by year. This provided a
temporal and spatial visualization of where different types of crime had occurred.
The third goal of the study was to look at how these crime types evolved over time for the
period of 2018-2020. The study accomplished this by dividing the crime hotspots into three time
periods. The first visualized hotspots that occurred in all three years, the second visualized
hotspots that occurred in two of the three years, and the third displayed hotspots that occurred in
just one of the three years. For a future study, a fourth section could show areas of the study
where no hotspots occurred.
The fourth goal of the study looked at the relationship between local polygons and point
features that may be related to different types of crime at different points in time. Point features
included bus stops and bars, both of which have been connected with crime in previous studies.
Polygons included single-family, multi-family, industrial, commercial, park and open space, and
undeveloped land uses. These relationships may point to explanatory features that help law
enforcement better understand which areas are particularly prone to crime and thus provide more
information to help in their fight to combat crime.
7

The fifth goal of the study was exploratory, and looked at whether there is a relationship
between different types and amounts of crime and natural disasters. Two tropical storms (Beta
and Imelda) and one major hurricane (Harvey) occurred before and during this 2018-2020 crime
study. The results of this analysis will enable law enforcement to better prepare for changes in
crime rates, especially during and following natural disasters when resources may be limited.
1.5 Thesis Structure

The remainder of this thesis consists of four chapters. Chapter 2 describes the related
work that informed this study. Past crime research including spatial analysis of crime, analytical
techniques, and the relationships between land use and crime are reviewed. Chapter 3 describes
the spatial analysis methods and techniques used for this thesis project, including the optimized
hotspot tool that was utilized to select land use attributes that fell within the highest crime
hotspots with a 99% confidence. Chapter 4 describes and discusses the results of the 2018-2020
hotspot analysis of four crime types in Houston, TX. Heat maps (otherwise known as density
maps) were created to show temporal patterns of crime. A number of maps are used to explain
the spatial and temporal patterns for each of the four crime types. Chapter 5 offers some
reflections as to the strengths and weaknesses of this project, the applicability of this work to
similar projects, and suggestions for future research.

8

Chapter 2 Related Work

This chapter describes the related work for the development of this 2018-2020 Houston, TX
crime trend study. Section 2.1 discusses the temporal and spatial extents of related studies and
section 2.2 explores whether there is a relationship between place, such as land use, and crime.
The final section of this chapter, section 2.3, describes how innovative methods were used in the
analysis of crime.
2.1 Temporal and Spatial Extents
Groff and LaVigne (2002) determined that scale and accuracy are important when
conducting crime hotspot studies, and that choosing the appropriate temporal and spatial scales
for these types of study is critically important. In general, a short time period, such as a month or
two, will not provide enough data to delineate accurate crime hotspots (Groff and LaVigne 2002;
Spelman 1995). Spelman (1995) suggested that police should use annual crime data instead of
weekly or monthly crime data to predict hotspots of crime with 90% accuracy. Groff and
LaVigne (2002) described how the geocoding of the crime addresses into coordinates can be
used to support spatial analysis and visualization of the crime patterns.
Groff and LaVigne (2002) also discussed choosing the appropriate spatial scale and how
the crime data can be aggregated by raster grid cells or police precincts. These authors suggested
using square raster grid cells that measured 500 ft on a side to perform crime hotspots in contrast
to Gorr and Olligschlaeger (2002), 4,000 ft on a side for optimum hotspot accuracy. The choice
of square grid cell size for this study was addressed to determine whether a custom setting should
be applied or an Esri default setting should be utilized for the hotspot raster grid cells. A study of
violent crime conducted by Esri in 2015 used a distance interval of 1,375 feet to determine
violent crime hotspots and an optimal fixed distance band of 4,556 feet. This particular study
9

suggested examining violent crime hotspots to public high schools (Esri 2015). The
abovementioned studies suggested examining different raster grid cell sizes for effective hotspot
analysis and using buffers to select certain land use types within a specified distance of the crime
hotspots. The use of square grid cells was advocated by Gorr and Olligschlaeger (2002) and
Groff and LaVigne (2002).
Groff and LaVigne (2002) also discussed the amount of data that is needed for an
accurate study of crime hotspots and suggested that it is important to choose an area that is large
enough to provide an adequate number of observations. Gorr and Olligschlaeger (2002) found
that a dataset of 30 or more crime occurrences per month provided reliable hotspot accuracy. The
study of violent crime conducted by Esri in 2015 had a dataset of over 22,000 occurrences of
crime for 2014.
2.2 Place and Crime
The existing literature also documents how certain land use types are related to crime hot
spots. Commercial areas, for example, contribute to certain types of crime (Block and Block
1995; Levine, Wachs and Shirazi 1986). Sherman, Gartin and Buerger (1989) showed that half
of the crime hot spots in Minneapolis, MN occurred near bars. Roncek and Maier (1991) also
discovered a correlation between high levels of crime and bars in Cleveland, OH.
Curry and Spergel (1988) argued that the location where a resident resides is a critical
factor in determining whether an individual will participate in criminal activity or not. These
authors found crime “to be correlated with poverty and a lack of social control, but violence
(e.g., homicide) to be correlated with their measure of social disorganization” (Curry and Spergel
1988, 218). Cohen, Gorr, and Olligschlaeger (1993) found that drug-related crime hot spots
occurred near bars, dilapidated commercial buildings, and/or areas associated with poverty.
10

Levine (2015) found that some types of crime and the associated hotspots occurred in
mixed land uses. For example, Canter (2006) demonstrated that more thefts occurred at bus and
light rail stations and Paisely (2017) also found that there was a relationship between crime
hotspots and distance to metro rails and bus stops.
2.3 The Use of Innovative Methods in Crime Analysis
Nakaya and Yano (2010) employed exploratory data analysis to visualize space time
kernel density estimation and scan statistics in Kyoto City during the time period of 2003-2004.
The authors used space-time statistical approaches to delineate crime clusters in space and time
and space-time cubes to visualize space and time point clustering in a 3D environment. The latter
approach displayed where crime hotspots occurred and how long they lasted. Old, persistent,
emerging, and new hotspots were delineated utilizing space-time kernel density estimation
(STKDE) and space-time scan statistics (STSSS) (Nakaya and Yano 2010). The results of this
study showed crime hotspots in the center of the city and around the Kyoto as well as other
railway stations. The criminal activities favored areas where criminals had found success in the
past, which were labeled as “known crime scenes” (Nakaya and Yano 2010, 237). These authors
also suggested documenting crime occurrences on a weekly and/or daily basis.
Another study performed a space-time analysis of crime (STAC) focusing on the
downtown area of the Buffalo, NY metropolitan area (Shiode 2011). Numerous authors
examined crime hotspot differences at the street level using the Street-level Spatial Scan Statistic
and STAC for analyzing street crime occurrences. The aforementioned study used spatial and
temporal techniques along with “spatial scan statistics” to detect crime hotspots (Shiode 2011,
365). This method was tested by simulating crime points at the street level which was then
compared with street level crime data, focusing on drug-related crimes and robberies in
11

downtown Buffalo, NY in the mid-1990s (Shiode 2011). The results showed crime trends for
both crime types by documenting how the hotspot locations changes over time. Some crime
locations remained the same for two years and some new hotspots emerged in 1996. Shiode
(2011) determined that the network spatial and temporal analysis of crime (NT-STAC) and
network spatial and temporal scan (NT-SaTScan) statistics provided more capable methods for
finding hotspots at the street-level. This new method was an adaptation of spatial scan statistics
and the classic spatial and temporal analysis of crime and the simulations supported the idea that
these two methods are more valuable in finding street crime hotspots than previous methods.
Another study by Li et al. (2018) that focused on earthquakes provided guidance on how
to handle hotspots with varying geographic extents. This author explored a new method for
finding clusters of occurrences that varied over time and space by identifying occurrences that
fall within prominent hotspots. Two parameters were suggested for clustering. The first
measured the hotspot mass with respect to space and time, and the second created a mass
tolerance for determining important hotspot clusters. Li et al. (2018) used the spatio-temporal
density based scan (ST-DBSCAN), windowed nearest neighbor (WNN), and spatio-temporal
shared nearest neighbors (STSNN) approaches to find hotspots that had different masses and
sizes and concluded that the newer, proposed spatio-temporal window (STW) technique was
more successful in finding hotspots with different forms and masses than previous ones. Li et al.
(2018, 324) used the term “clusters” to describe earthquake hotspots and this same term was used
to define crime hotspots in this study.
Another recent study by Wheeler et al. (2018) introduced two techniques for measuring
contrasts in point patterns. The first method analyzed several point patterns and found the
discrepancies in the amount of point clustering. When the discrepancies were found, number
12

breaks could be created from the changes in percentages. This method determined whether or not
the discrepancies in the amount of point clustering was important. The second method used
regression techniques to determine changes in the size of point patterns over a time period.
Wheeler (2018) used these methods to compare the location of where the NYPD pedestrian stops
occur and where violent crimes occurred between 2006 and 2016. Two point patterns were
statistically analyzed using grid cells to determine the correlation in the number of pedestrian
stops and the number of violent crimes. The results showed how the violent crime and police
stops remained steady between January 6
th
and January 13
th
2012 and that over the next three
days, January 14
th
- January 16
th
2012, the number of police stops dramatically increased and the
number of violent crimes trended downward (Wheeler et al. 2018).
The results provided in the Wheeler et al. (2018) demonstrated how knowledge of crime
hotspot location could permit the police to reduce the amount of crime in a short period of time
by reallocating law enforcement resources. This also means that retrospective studies like this
one would benefit tremendously from knowledge of how law enforcement resources were
allocated over the period of study. However, these data are seldom available and this makes it
difficult to interpret the driving forces for changes in crime patterns.

13

Chapter 3 Research Design and Methods

Crime data and study location were key factors for this thesis project. The HPD provided crime
address data that was geocoded into x,y coordinates. The inner I-610 region of Houston, TX
(Figure 1) contained greater than 100,000 crime incidents for the 2018-2020 period. This data
was transformed into crime density maps to reveal crime trends and patterns.
The first step of this project was finding the crime data and deciding how to organize it in
Excel. The next step was to determine if the crime points had adequate address information.
Once crime point address information was confirmed, geocoding was accomplished using the
Esri ArcGIS Pro Geocoding Service, CDX Winzip Geocoder in Excel and Geocoding by Smart
Monkey in Google sheets. The geocoded crime points were added into ArcGIS Pro using the
display x, y coordinates tool to facilitate analysis and visualization.
ArcGIS supports a number of methods for visualization and analysis of crime hot spots
and demographics. Within the Esri workspace, the Kernel Density and Optimized Hot Spot
Analysis tools were chosen for this study. Once the heat maps were created using the Kernel
Density tool, the hotspots were classified into five categories. The Optimized Hotspot tool
produced crime hotspots with 99% confidence for this study.
The city of Houston GIS (COHGIS) provided a land use layer with ten different land use
types and a bus stop point layer. The land use layer was overlaid on the optimized hotspot areas,
and the bus stop point shapefile provided the locations of all of the bus stops within the study
area. Both of these datasets were used to explore associations with high crime areas.
The research design is summarized below in Table 1. The workflow used for data
acquisition, pre-processing, analysis, and mapping is listed in Table 1 and the six subsections
which follow describe the methods and data sources in more detail.
14

Table 1. Thesis Project Workflow
Fourteen Steps
1. Acquired crime incident dataset from HPD
2. Downloaded data as Excel Spreadsheets
3. Geocoded addresses into latitude and longitude coordinates
4. Organized data spreadsheets by location (police beats), date, and crime type
5. Added crime spreadsheets into ArcGIS Pro as database tables
6. Displayed x, y coordinates as crime points
7. Performed Kernel Density Analysis on crime data
8. Classified the crime hotspots into five categories
9. Performed Optimized Hot Spot Analysis on crime data
10. Mapped crime hotspots
11. Added COH land use layer
12. Added bus stations and bar locations
13. Created multiple maps using a combination of different data layers
14. Exported the maps to 750 dpi PDF format and created final versions of the map

3.1 Description of Crime Attributes
The crime data was downloaded as Excel spreadsheets from the Houston Police
Department Crime Statistics for 2018, 2019, and 2020. These data can be downloaded at
https://www.houstontx.gov/police/cs/Monthly_Crime_Data_by_Street_and_Police_Beat.htm.
The 2019 and 2020 crime data was downloaded for the entire year and the 2018 data were
downloaded by month.
The crime points were provided as rows in Excel spreadsheets with associated addresses
(street numbers and names), type of crime, and crime date. There were 100,791 total crime
occurrences in this dataset. The Excel spreadsheets were converted into CSV (comma-separated
value format), tables and were then added into ArcGIS Pro as database tables.
The attributes accompanying each of the crime points included incident number,
occurrence date, occurrence hour, crime offense type, beat, premise, block range, street name, type
of street, city, state, and ZIP code. The HPD switched to the National-Incident Based Reporting
System (NIBRS) in June 2018 and prior to June 2018, the HPD used the Uniform Crime Reporting
15

system. The pre-June 2018 attributes were date, hour, crime offense type, beat, premise, block
range, street name, and type of street. The CDX WinZip tool in Excel was used to find ZIP codes
for each of the pre-June 2018 crime data points using block number, street name, city, and state.
The CDX Locate Bing Function was used to find pre-June 2018 ZIP codes. Three columns in
Excel were created for the pre-June 2018 spreadsheets that included city, state, and CDX WinZip
located ZIP code.
3.2 Organization of Crime Data
The HPD crime data was organized by police beats, also known as police jurisdictions.
The police beat data was obtained from the HPD Crime Statistics (City of Houston Police
Department 2021 a and b). A filter was run on the police beat column to select the 48 police
jurisdictions located within the study area. A second filter was used to select the four crime types
for this study and another filter was used to separate the yearly data into monthly data.
The four crime types varied depending on the database used. The pre-June 2018 crime
types were Assault (aggravated assault), Burglary (burglary), Robbery (robbery), and Theft,
(auto theft and theft). The post-June 2018 crime types were Assault (aggravated assault, simple
assault), Burglary (burglary, breaking and entering, burglary), Robbery (pocket-picking, purse-
snatching, robbery), and Theft (stolen property offenses, theft from buildings, theft from motor
vehicles, and theft of motor vehicle parts or accessories). These later crimes were collapsed to
match the pre-June 2018 crime types, as discussed in the next paragraph.
Crime types were aggregated for analysis. For example, all theft categories were
combined into a single category whether there were two attribute categories (pre-June 2018) or
four attribute categories (post-June 2018). Aggravated assault and simple assault were
aggregated. This study matched the post-June 2018 crime types with the pre-June 2018 crime
16

types. The 2019 and 2020 crime data were merged into one Excel spreadsheet for each year. The
pre-June 2018 data was organized by month with the data for the remainder of 2018 combined
into one annual spreadsheet. SQL was used to filter crime types in ArcGIS Pro.
3.3 Geocoding of Crime Addresses
The crime point data was downloaded from the HPD for all three years, 2018-2020. The
2020 crime data was geocoded using Esri’s World Geocoding Service. The 2018 and 2019 crime
data was geocoded using the CDX WinZip geocoder tool in Excel. These tools turned street
addresses into latitude and longitude coordinates. Geocoding by Smart Monkey in Google sheets
was utilized to geocode the 2% of crime addresses that could not be located with Esri World
Geocoding Service and/or CDX WinZip.
3.3.1 Geocoding in Excel using CDX WinZip
The CDX WinZip geocoder was downloaded for Excel as an add on. The address
provided by the HPD for the 2018 and 2019 crime data was divided into five columns: street
number, street name, city, state, and zip code, and a formula was created in Excel to merge these
five columns into one that showed the full address.
The Insert CDX LocateBing Function was opened and a single line address was entered
for a selected row and column. The copy function was used for the entire column, which allowed
3,000-5,000 monthly crime events to be geocoded at a time using the executive formula button.
This first run produced geocoded addresses with a high level of confidence using Bing Maps.
The Bing maps tool specifies accuracy as High, Medium or Low confidence and the 2% of the
addresses that could not found with high confidence were not returned with latitude and
longitude coordinates. A column was then created to display the geocoding confidence level for
each crime address. This confidence level provided an accuracy assessment of the geocoded
17

location of the given latitude and longitude coordinates. The next paragraph describes the
methods used to handle the second run of crime occurrence addresses.
The addresses that were not found on the first run were rerun, this time with the Allow
Ambiguous Data box checked. This function allowed addresses to be geocoded with medium
confidence. With this second run, almost all (i.e. 95-98%) of the crime addresses for 2018 and
2019 were located using Bing Maps. This also generated latitude and longitude in the same
column, which was problematic for future analysis. Latitude and longitude needed to be in two
separate columns in Excel for use with the ArcGIS Pro XY Table to Point tool. The Excel text to
columns tool was used to separate the combined latitude and longitude columns. The 2-5% of the
crime addresses that were not located with the CDX Winzip geocoder used another geocoder in
Google Sheets. The Geocoding by Smart Monkey tool is discussed in the next section.
3.3.2 Geocoding in Google Sheets using Geocoding by Smart Monkey
For the addresses that were not found during the first and second geocoding attempts in
CDX WinZip, the Geocoding by Smart Monkey application was downloaded in Google Sheets.
Geocoding by Smart Monkey was used to try and find the 2% of the CDX WinZip location
points not found. A template was created for street number, street name, city, state, and ZIP
code. The unlocated geocoding addresses were added into the Geocoding by Smart Monkey
template and this approach was able to find almost all of the addresses that were not successfully
geocoded on the first and second runs of the CDX WinZip tool.
The two CDX WinZip runs and Geocoding by Smart Monkey were utilized for all the
2018 and 2019 crime points and the results were merged to produce a master spreadsheet for
each year. The 2020 crime points were all geocoded using Esri’s World Geocoding Service. The
next section briefly describes this geocoding method using the Esri application in ArcGIS Pro.
18

3.3.3 Geocoding in ArcGIS Pro
The 2020 crime was geocoded using the ArcGIS World Geocoding Service. This is the
third geocoding service used in this study. Three different geocoding tools were used in this
study as a test of geocoding effectiveness. The ArcGIS World Geocoding Service was the most
time-efficient of all the geocoding methods used for this thesis project. The addresses were
imported as a single-line address format and the Esri World Geocoding Service was used to
convert addresses into x and y coordinates.
This study also used the Esri World Geocoding Service to geocode the crime addresses
for July-December 2020 that fell on the outer boundaries of I-610. These crime addresses were
downloaded from the HPD and were utilized to evaluate crime edge effects. This added an
additional 13,944 crime occurrences to the study data set and another 17 police precincts to the
study area. The Esri World Geocoding Service was used to geocode all of these crime addresses,
with 2.8% of the crime points being identified as in the same location.
3.4. Spatial Analysis of the Crime Data
The 2018 and 2019 master spreadsheets were imported into ArcGIS Pro as standalone
tables. The display x, y data tool was selected by using the XY Table to Point geoprocessing
tool. The coordinate system was GCS_WGS_1984 W. This projection was chosen because it
matched the COHGIS street layer and the COHGIS land use layer.
The Kernel Density and Optimized Hot Spot Analysis methods were selected to analyze
the crime data in Houston, TX. The Kernel Density tool was used to determine crime hotspots
for the period 2018-2020. The crime points were aggregated by using counts within raster grid
cells using the fishnet grid method for both the kernel density and the optimized hot spot analysis
methods.
19

The Kernel Density method produced interval value densities of crime counts. The
interval values offer a range of numbers and specify the confidence of a hotspot. Taking the 2019
Robbery dataset with 2,810 crime points as an example, the interval value range was < 728,436
and > 145,687. These interval values were classified into five hotspot categories: Very High,
High, Medium, Low and Very Low crime densities. For the 2019 Robbery dataset, these
densities were represented by 728,436, 582,749, 437,062, 291,374, and 145,687, respectively.
The study focused on the very high, high and medium hotspots, but some low hotspots are
mentioned in the results section. The very low hotspot categories were not considered further in
the study because they were seen as the least necessary with respect to the priorities of police, the
target audience for this study.
Esri’s default settings were used with the Kernel Density and Optimized Hotspot
Analysis tools to execute and delineate crime hotspots. The settings used by Esri (2015) did not
work for all the kernel density and optimized hot spot maps in this Houston crime study. The
Esri (2015) used a small number of crime occurrences compared to this Houston crime study.
This Houston crime study volume used tens of thousands of occurrences of crime for
each year and crime type. Therefore, the Esri default settings were able to determine hotspot
densities even when there were a large volume of crime occurrences concentrated in close
proximity to each other. The trials using the customized settings generated an error message.
This study also used Esri default settings to determine raster grid sizes because these settings
accounted for the largest numbers of crime occurrences.
The Optimized Hot Spot Analysis method implemented in ArcGIS Pro determined the
locations and accompanying probabilities of crime hotspots, with Gi Bin Level 3 statistics
suggesting a 99% chance of a hotspot. Parameters were set by using the Esri default setting’s
20

grid cell size and distance bands. The use of Esri’s customized settings created crime density
maps from the HPD geocoded crime points.
Using crime occurrence incidences, the optimized hot spot analysis tool produced a map
with high statistical confidence of hot and cold spots using ArcGIS Pro’s Getis-Ord Gi statistics.
The Gi_Bin statistics made up of z-scores and p values show the significance of hot and cold
spots. The z-scores represent standard deviations and the p values show the probability and
confidence levels. The Gi Bin z-scores are either +3 or -3 and reflect the 99% confidence level,
whereas the +2 or -2 bins reflect the 95% confidence level, and the +1 or -1 bins reflect the 90%
confidence level (Esri 2021). The negative (-) and positive (+) signs represent cold and hot spots,
respectively. This study used the Gi Bin level (+) 3 to delineate crime occurrence hotspots at the
99% confidence level.
3.5 Land Use and Optimized Hotspots
The COH land use layer, downloaded from the City of Houston, was clipped so that only
attributes within the study area were visible and projected in WGS 84. There were 205,711
individual land use attributes that fell within the study area representing 10 land-use categories.
They are Agriculture Production, Commercial, Industrial, Multi-Family, Office, Park and Open
Space, Public and Institutional, Single-Family, Transportation and Utility, and Undeveloped
Land Use. The acreages for each land use type were recorded in this dataset.
The land use types that fell within these Gi Bin Level 3 crime hotspots were selected for
further analysis. The select by location query tool was used to select these land use attributes
within the polygon (boundary) of the Gi Bin Level 3 crime hotspots. The land use statistics
provided percentages of what type of land use was most prevalent in each optimized crime
hotspot location for each type of crime.
21

The optimized hotspot tool was conducted on all four crime types for 2020 and given the
land uses included in the areas with hotspots, the study focused on Commercial, Industrial,
Multi-Family, Single-Family, Park and Open Space, and Undeveloped land uses. Kernel density
analysis was conducted on the bus stop point layer, downloaded from the City of Houston, to
determine the density hotspots of bus locations. The results of this bus stop kernel density
analysis is displayed in Appendix C. Appendix A shows bus stop density and the location of
2020 Theft hotspots. The final part of the analysis examined the associations between
commercial land uses, such as bars, and bus stop locations with the crime hotspots. The bar
locations within the study districts are shown in Appendix B. The results of these findings will
be discussed in the next chapter.

22
Chapter 4 Results

Table 2 lists the 2018-2020 crime statistics and shows that thefts and assaults exceeded
burglaries and robberies in all three years by considerable margins. The overall totals point to
more than 40,000 crimes with higher numbers in 2019 compared to the two years that proceeded
this particular year.
Table 2. 2018-2020 Crime Statistics

Year Assault Burglary Robbery Theft
2018 8,032 4,417 2,224 12,597
2019 12,385 5,239 2,810 15,954
2020 12,013 4,463 2,366 11,926

Totals: 32,430 14,119 7,400 40,477

The first section below looks at the location of the crime type locations within the study
area. The hotspots were categorized using 5 classes. The classes are very high (5), high (4),
medium (3), low (2), and very low (1). The study focused on the very high and high hotspot
crime types. The medium hotspot areas are also important because these areas could become
problematic in the future.
4.1 Crime Trends and Results of all Crime Type hotspots
4.1.1 2018-2020 Assault Hotspots
The 2018 category 5 and category 4 assault hotspots occurred mainly in the midtown-
10H40 and downtown 1A10 districts (Figure 3a). A large category 5 and category 4 hotspot of
2018 Assault occurred in the western to northwestern portion of the downtown district. This was
further north than the 2019 and 2020 category 5 and category 4 assaults (Figures 3b and 3c). This
category 4 and category 5 hotspot extends south by a narrow category 4 hotspot to another
category 5 and category 4 hotspot cluster. This category 5 and category 4 hotspot is located in
23
the northern portion of the midtown district near I-45 (Figure 3a). This category 5 and category 4
hotspot in 2018 is near the same location as the 2020 category 5 and category 4 cluster and just
east of the two 2019 category 5 and category 4 2019 assault hotspot clusters (Figures 3b and 3c).
In 2018, the remaining and majority of the remainder of the downtown district
experienced category 3 and category 2 hotspots. The category 2 hotspots in 2018 was smaller
than in 2019 and 2020. The category 2 cluster was located in the downtown and midtown
districts in 2018. In 2018, the Montrose district has a small cluster of category 2 in the central
portion of the district. In 2019, the category 2 extended across the Montrose district and into the
Kirby District. In 2020, this category 2 diminished to the intersection of Westheimer and Lincoln
where a small category 3 occurred.
In 2019, a larger category 5 and category 4 hotspot occurred at the intersection of the
Montrose district 1A20, Midtown 10H40, and Downtown 1A10 districts on I-45 near the same
location in 2020 (Figure 3b). This large problematic assault hotspot concentration is centered at
the northwest portion of the midtown district. In 2019, another category 5 and category 4 cluster
emerged in the central Montrose district that was not present in 2018 and 2020. This category 5
and category 4 in the downtown and midtown districts were connected by a narrow category 4
hotspot in 1A20 Montrose in 2019 (Figure 3b). The category 3 in 2019 covered most of the
Montrose, midtown, and downtown districts and was much larger than in 2020 (Figure 3c). The
category 3 was confined to the downtown and midtown districts in 2018 and 2020. This category
3 cluster extended into the center of the Montrose district in 2019. The three parts of Figure 3
show how assault dramatically expanded from 2018 to 2019 in the downtown and midtown
districts and then diminished in area in 2020. The location of the downtown, midtown and
Montrose districts are displayed in Figure 2. The police jurisdictions are shown in Figure 1.
0 3 0 6 0 1 5 M ile s
±
H a r r is C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
( a ) A s s a u l t H o t s p o t s i n 2 0 1 8 ( b ) A s s a u l t H o t s p o t s i n 2 0 1 9
F i g u r e 3 . 2 0 1 8 - 2 0 2 0 A s s a u l t C r i m e H o t s p o t s
2 4
( c ) A s s a u l t H o t s p o t s i n 2 0 2 0
2 0 1 8 A S S A U L T
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
2 0 1 9 A S S A U L T
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
2 0 2 0 A S S A U L T
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
4 5
6 1 0
6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
4 5
6 1 0
6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
4 5
6 1 0
6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
1 in c h e q u a ls 4 .5 m ile s
1 in c h e q u a ls 4 .5 m ile s
1 in c h e q u a ls 4 .5 m ile s
25
In 2019, the majority of the downtown, midtown and Montrose districts experienced a
large category 2 cluster. This category 2 cluster fully covered police jurisdictions 10H60, 10H50,
10H30, 1A10, 1A20, 10H40, 10H70, and 1A30 in 2019. A very large category 2 expanded from
the Kirby District to the Montrose District and covered the majority of the midtown and
downtown districts in 2019. In 2018, this category 2 was just a cluster at Westheimer and
Lincoln (Figure 3a.). In 2020, the category did not extend past Westheimer and Lincoln in the
Montrose district (Figure 3c).
In 2019, another category 5 and category 4 cluster emerged in the central Montrose
district that was not present in 2020 at the intersection of Westheimer and Lincoln. This category
5 and category 4 cluster connects to the large category 5 and category 4 in the downtown and
midtown districts by a narrow category 4 hotspot in 1A20- Montrose. The category 3 cluster in
2019 covered most of the Montrose, midtown, and downtown districts and was much larger than
in 2020. The category 3 cluster extended into the center of the Montrose district.
Category 5 and 4 assault hotspots also occurred near the intersection of the midtown 10H40
and downtown 1A10 districts on I-45. This category 4 and 5 hotspot is surrounded by a
category 3 hotspot in 10H40, 1A10 and 1A20 Montrose. The remaining areas of the downtown
and midtown saw category 2 assault hotspots. The Montrose District has a small category 3
cluster surrounded by a large area of category 2 hotspots. The category 3 cluster in the
downtown and midtown district extends 2.48 miles from the southwest to the northeast. From
the northwest to the southeast, the longest point is 1.2 miles across.
The other category 3 clusters in 2020 occurred at Sampson and Elgin, at the intersection
of Cavalcade and Lockwood and Liberty and I-69. These category 3 clusters are surrounded by a
large category 2 cluster. In 2019, the category 3 clusters disappear and six small category 2
26
clusters pop up in the northeast section of the study. A category 3 cluster occurred at Kirby and
I-69 in 1A30 in 2019. In 2020, there were no hotspots of assault in any category at Kirby and
the I-69.
In 2018 and 2020, there were no category hotspots that occurred along Washington
Avenue. In 2019, a new category 4 and category 3 cluster emerged on Washington Avenue
centered at Durham and Allen in 2A50. A large category 2 assault cluster surrounded this
category 4 and category 3 cluster along Washington Avenue in 2019. In 2020, there were no
assault hotspots along Washington Avenue.
In 2018, a small category 3 cluster existed at Scott and Amos in the southern section. In
2019, this small category 3 diminished to a category 2 cluster. Then in 2020, this cluster
remerged as a medium sized category 4 surrounded by a category 3 cluster. 2018 also saw a
large category 2 cluster from Scott to the west and MLK to the northeast of I-610. Medium sized
category 2 clusters occurred at Sunbeam and Scott and Wilmington and Cullen. The category 2
cluster ran north to south along Cullen south to Sunbeam in 2018. Another large category 2
cluster occurred at Bellfort and MLK. In 2018, there were only category 2 clusters in the
southern section except for a small category 3 cluster at Scott and Amos.
In 2020 in the southern portion of the study east of the intersection of 288 and I-610, a
small category 4 hotspot occurred in 14D20 and 14D10 in 2020, at the intersection of Scott and
Corder. It is surrounded by a cluster of category 3 hotspot.
There are five other category 3 clusters in the southern section in 2020. One occurs near
Yellowstone and Cullen, and the others occur near Milart and MLK, Cullen and I-610, Sunbeam
and Scott and Sunbeam and Cullen. A large category 2 cluster completely covers police beats
14D20, 14D30, 14D10 and 10H60. The category 2 cluster located east of 288 and north of I-610
ran east to west 2.86 miles and 1.95 miles north to south. South of I-610 and east of 288, this
27
category 2 ran 3.72 miles from the southwest to the northeast and 2.43 miles from north to south
in 2020.
In 2019, the category 4 and category 3 cluster at Scott and Corder diminished to a small
category 2 hotspot. The other two category 2 s north of I-610 also diminished to a medium sized
category 2. The large category 2 in 2020 covering all of 14D20, 14D30, 14D10 and 10H60
dissipated into small category 2 clusters at Sunbeam and Scott and Sunbeam and Cullen. In both
2019 and 2020, a medium sized category 3 and 2 cluster occurred at 90 and I-610. All three
years saw a category 2 cluster at Canal and Wayside.
At Fulton and Morris, a category 2 cluster existed in both 2019 and 2020. At I-10 and 69
in 2018, a category 2 hotspot occurs north of I-10. The category 2 is east of 69 and spreads up
and along 69 from the north to the south in 7C10 and 7C20. This large category 2 southern
border is I-10, 69 to the east and Crane to the north. East of I-69, the western portion runs 1.89
miles north to south down Gregg and 0.82 miles from west to east along Crane in 7C20 in 2018.
This large category 2 hotspot along Crane occurs from I-69 to the west and Hoffman to the east
in 7C20 in 2018.
Figure 4 shows the assault hotspots that occurred in just one year, in two of the three
years and hotspots that occurred in all three years. Figures 4(a) and 4(b) show how assault in
2019 was much more expansive in the downtown and midtown districts than in 2018 and 2020.
Figures (4 c-f) show the 2019 assault hotspots that only occurred in specific areas of the study in
that one year. Figures (4 g-i) show the northern, southern and southwestern areas of the study
where hotspots occurred in all three years. Figures (4 j-m) display the southern and southeastern
areas of the study area where hotspots occurred in two of the three years.
( g ) - ( i) A s s a u lt H o t s p o t s in a ll 3 y e a r s
I - 6 9
C a v a lc a d e
G r e g g
L ib e r t y
A lm e d a
S c o t t
M L K
Y e llo w s t o n e
G r ig g s
B e llf o r t
S u n b e a m
C u lle n
M L K
C a n a l
B e llf o r t
C u lle n
W a s h in g t o n
H o lc o m b e
M L K
B e llf o r t
W a s h in g t o n
A lm e d a
( a ) O v e r la y o f 2 0 1 8 a n d 2 0 2 0 A s s a u lt H o t s p o t s
( b ) 2 0 1 9 A s s a u lt H o t s p o t s
( a ) a n d ( b ) 2 0 1 8 - 2 0 2 0 A s s a u lt H o t s p o t s in a ll 3 y e a r s
( g ) 2 0 1 8 - 2 0 2 0 c a t e g o r y 3 a n d c a t e g o r y 2 h o t s p o t s
b e t w e e n C a v a lc a d e a n d I - 1 0 .
( h ) 2 0 1 8 - 2 0 2 0 c a t e g o r y 4 , 3 a n d 2 h o t s p o t s in t h e
s o u t h e r n s e c t io n .
( i) 2 0 1 8 - 2 0 2 0 c a t e g o r y 3 a n d c a t e g o r y 2 h o t s p o t c lu s t e r
a t t h e in t e r s e c t io n o f 9 0 a n d I - 6 1 0 .
( c ) 2 0 1 9 c a t e g o r y 4 , c a t e g o r y 3 a n d c a t e g o r y 2 h o t s p o t
c lu s t e r a lo n g W a s h in g t o n A v e n u e .
( d ) C a t e g o r y 3 a n d c a t e g o r y 2 h o t s p o t c lu s t e r a t K ir b y a n d I - 6 9 .
T h e r e is a n o t h e r s m a ll c a t e g o r y 2 c lu s t e r o n W e s t h e im e r
n e a r I - 6 1 0 .
( e ) 2 0 1 9 c a t e g o r y 2 c lu s t e r a lo n g 1 9 t h b e t w e e n
B e v is a n d A s h la n d in t h e N W s e c t io n . ( f) 2 0 1 9 c a t e g o r y 2 a t I - 6 1 0 a n d A lm e d a .
( j) 2 0 1 8 - 2 0 2 0 c a t e g o r y 2 h o t s p o t s s o u t h o f I - 6 1 0 .
( l) 2 0 1 8 a n d 2 0 1 9 c a t e g o r y 2 a s s a u lt h o t s p o t
a t C a n a l a n d 9 0 .
( m ) 2 0 1 9 a n d 2 0 2 0 c a t e g o r y 2 a lo n g H o lc o m b e .
I - 1 0
9 0
I - 6 1 0
9 0
I - 6 1 0
9 0
I - 6 1 0
I - 4 5
A s h l a n d
1 9 t h
B e v i s
W e s t h e i m e r
K ir b y
K ir b y
( c ) - ( f) 2 0 1 9 A s s a u lt H o t s p o t s ( j) - ( m ) 2 0 1 8 - 2 0 2 0 A s s a u lt H o t s p o t s in 2 y e a r s
F ig u r e 4 . F r e q u e n c y o f A s s a u lt H o t s p o t s
( k ) S E s e c t io n h o t s p o t s f o r 2 0 1 9 a n d 2 0 2 0 .
1 in c h e q u a ls 2 .5 m ile s
1 in c h e q u a ls 2 .5 m ile s
1 in c h e q u a ls 2 .5 m ile s
1 in c h e q u a ls 1 .5 m ile s
1 in c h e q u a ls 2 .5 m ile
1 in c h e q u a ls 3 m ile s
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 2 m ile s 1 in c h e q u a ls 3 m ile s
1 in c h e q u a ls 2 m ile s 1 in c h e q u a ls 1 .5 m ile s
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 4 m ile s
± ±
± ± ± ±
± ±
±
± ± ±
±
S u n b e a m
9 0
I - 6 1 0
I - 6 9
C u lle n
9 0
9 0
2 0 1 9 A s s a u l t
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
2 0 2 0 A s s a u l t
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
2 0 1 8 A s s a u l t
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
0 3 0 6 0 1 5 M i l e s
±
2 8 8
L o c k w o o d
H A R R IS C O U N T Y , T X A N D S T U D Y A R E A (O R A N G E )
2 8
29
In the northeast portion of the study area north of I-10 and east of 69 in 2019, there are
small category 2 clusters. Three of these category 2 clusters are on I-10. The western category 2
is 0.45 by 0.58 miles in 7C10 at Gregg and I-10. The second one is 0.5 by 0.59 miles in 7C10 at
Lockwood and I-10. The eastern one is 0.5 by 0.56 miles in 9C20 at Harbor and I-10. A small
category 2 cluster occurs at Gregg and Liberty. At Lockwood and Cavalcade, a medium sized
category 2 exists. A very small category 2 cluster occurs at I-69 and Collingsworth. At
Lockwood and I-610, a very small category 2 cluster exists.
In the northwest portion of the study area, from west to east there is a category 2 hotspot
that extends from Bevis to Ashland in 2A60 and 2A30 in 2019. This category 2 cluster runs 1.45
miles east to west and 0.68 miles north to south. A large category 2 cluster occurs at MLK and
Bellfort with another large category 2 cluster between Scott and Cullen south of Bellfort in 2019.
A large area of category 2 (low) hotspots occurs northeast of the intersection of Interstate
10 and I-69 in 2020. One cluster of category 3 (medium) hotspots occurs along and just east of I-
69 and north of I-10. This category 3 cluster runs 0.66 by 0.76 miles. Another category 3 cluster
occurs at the intersection of Cavalcade and Lockwood and runs 0.62 by 0.9 miles.
A category 2 hotspot assault runs along I-10 from the intersection I-69 and I-10 to the
east for over 3 miles in 2020. This category 2 is 4.32 miles at the longest point from west to east
and 3.4 miles at the longest point from north to south. The western portion of this large category
2 hotspot is between I-45 to the west and I-69 to the east. The center of this category 2 is near
Fulton and Morris in 2020. In 2018, there are two medium sized category 2 clusters at Gregg
and Liberty and along Crane.
30
4.1.2 2018-2020 Summary of Burglary Hotspots
The 2018 burglaries occurred mainly in the midtown and downtown districts (Figure 6).
Just east of the intersection of Allen and 45 in the downtown district, there was a large category
5 and 4 cluster of burglary in the downtown district. This category 5 and category 4 hotspot was
north of the 2019 and 2020 category 5 and category 4 hotspots along the western border of
downtown. The category 3 extends from the southwest portion of the midtown district along the
western borders of the midtown and downtown districts all the way to the northwest portion of
the downtown district. The remaining portions of the midtown district saw a category 2 hotspot.
The eastern, southeastern, and central portions of the downtown district experienced a category
2 hotspot. The central to eastern half of the Montrose district is a category 2 hotspot. The
eastern Montrose border area saw a category 3 hotspot in 2018. The majority of the Kirby
district experiences a category 2 hotspot in 2019.
A larger category 5 and 4 burglary hotspot occurred near the intersection of the Montrose
1A20, Downtown 1A10 and Midtown 10H40 districts in 2019. This category 5 hotspot was
located in the northwest portion of Midtown and the eastern portion of the Montrose district,
near the intersection of Brazos and I-45. The central and eastern portion of the Montrose District
in 2019 saw a large category 5 and category 4 hotspot that was part of the same large category 5
and category 4 hotspot in the northwestern midtown district. The majority and/or entire
Montrose, Downtown and Midtown districts was covered in category 3 and category 2 hotspots
for 2019. This category 5 and 4 hotspot grew and expanded into Montrose from 2020 to 2019.
The large category 2 hotspot was present in both 2020 and 2019 and both were much larger in
2019.
Several category 5 and 4 burglary hotspots occurred near the intersection of the
downtown 1A10 and midtown 1040 districts near Louisiana and I-45 in 2020. One category 5
0 3 0 6 0 1 5 M ile s
±
H a r r is C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
( a ) B u r g l a r y H o t s p o t s i n 2 0 1 8 ( b ) B u r g l a r y H o t s p o t s i n 2 0 1 9
F i g u r e 5 . 2 0 1 8 - 2 0 2 0 B u r g l a r y C r i m e H o t s p o t s
3 1
( c ) B u r g l a r y H o t s p o t s i n 2 0 2 0
2 0 1 8 B U R G L A R Y
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
2 0 1 9 B U R G L A R Y
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
2 0 2 0 B U R G L A R Y
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
4 5
6 1 0
6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
4 5
6 1 0
6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
4 5
6 1 0
6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 4 m ile s
32
and category 4 hotspot occurred near the same location as the category 5 and category 4 cluster
in 2019. A substantial category 3 hotspot surrounded this category 4 and 5 hotspot and covered
portions of the downtown, midtown, and Montrose districts. This category 4 and 5 hotspot was
located in the northwest portion of the midtown district. A very large category 2 hotspot covered
the remaining southwest portions of midtown, southern downtown, the majority of Montrose and
the Kirby district.
There was a small category 3 cluster at Shepherd and I-69 in 1A30 in 2020. A small
cluster of category 4 and 3 occurred near Kirby and I-69 in 1-30 in 2019. In 2018, there was just
one small category 2 cluster at Kirby and I-69. A large category 4 and 3 hotspot occurred along
Washington avenue at Durham in 2A50 in 2019. An expansive category 2 hotspot occurred
along Washington avenue from Westcott to the west and Studemont to the east in 2019. This
category 2 hotspot in 2019 was so large that it connected to the Montrose District category 5 and
category 4 hotspots to the south on Studemont Street.
In 2020, Washington Ave saw only two small category 2 clusters. In 2018, Washington
Ave had just one medium sized and one small category 2 cluster. Burglaries were much more
prevalent on Washington Ave in 2019. There was a category 2 cluster between TC Jester and
Thompson along Washington Avenue in 2018. At Washington Avenue and the I-45, a category 2
hotspot extends out to the north from the large category 5 and category 4 hotspot at Allen and I-
45 in 2018.
Figure 5 summarizes the 2018-2020 burglary crime trends. All three years showed a
large category 5 and category 4 hotspot cluster on I-45 near the intersection of the downtown
and midtown districts. From 2018 to 2019, this large category 4 and category 4 cluster expanded
into the eastern Montrose District and then diminished to a smaller size in 2020. Figure 6
summarizes the frequency of burglary crime trends.
( a ) - ( d ) 2 0 1 8 - 2 0 2 0 B u r g la r y H o t s p o t s in a ll 3 y e a r s
F ig u r e 6 . F r e q u e n c y o f B u r g la r y C r im e H o t s p o t s
( a ) 2 0 1 8 a n d 2 0 2 0 B u r g la r y H o t s p o t s
W a s h in g t o n
M o r r is
A lm e d a
S c o t t
M L K
Y e llo w s t o n e
G r ig g s
B e llf o r t
S u n b e a m
C u lle n
M L K
B e llf o r t
C u lle n
F u lt o n
C o llin g s w o r t h
L o c k w o o d
( h ) 2 0 1 8 - 2 0 2 0 c a t e g o r y 2 h o t s p o t s n e a r t h e in t e r s e c t io n
o f L o c k w o o d a n d C o llin g s w o r t h .
( i) 2 0 1 8 - 2 0 2 0 c a t e g o r y 2 h o t s p o t s in t h e s o u t h e r n s e c t io n .
( j) 2 0 1 8 - 2 0 2 0 c a t e g o r y 3 a n d c a t e g o r y 2 h o t s p o t
c lu s t e r a t t h e in t e r s e c t io n o f 9 0 a n d I - 6 1 0 .
( h ) - ( k ) 2 0 1 8 - 2 0 2 0 B u r g la r y H o t s p o t s in a ll 3 y e a r s
( k ) 2 0 1 9 c a t e g o r y 4 , c a t e g o r y 3 a n d c a t e g o r y 2 h o t s p o t
c lu s t e r w it h 2 0 1 8 a n d 2 0 2 0 c a t e g o r y 2 c lu s t e r s a lo n g W a s h in g t o n A v e .
( l) 2 0 1 9 a n d 2 0 2 0 B u r g la r y H o t s p o t s in t h e S E s e c t io n . ( m ) 2 0 1 8 a n d 2 0 1 9 B u r g la r y H o t s p o t s a lo n g I - 1 0 n e a r 9 0 .
( n ) 2 0 1 8 a n d 2 0 1 9 B u r g la r y H o t s p o t s a t F u lt o n a n d M o r r is .
( l) - ( n ) B u r g la r y h o t s p o t s in 2 y e a r s
( e ) 2 0 1 9 c a t e g o r y 2 's in t h e s o u t h c e n t r a l s e c t io n . ( f) 2 0 1 8 c a t e g o r y 2 in t h e s o u t h w e s t s e c t io n .
( g ) 2 0 1 8 c a t e g o r y 2 in t h e n o r t h e r n s e c t io n .
K ir b y
I - 6 1 0
9 0
I - 4 5
I - 6 1 0
I - 4 5
I - 6 1 0
I - 4 5
I - 6 9
I - 6 9
I - 1 0
I - 1 0
9 0
S u n b e a m
9 0
L ib e r t y
F u lt o n
C o llin g s w o r t h
I - 6 1 0
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 3 m ile s 1 in c h e q u a ls 3 m ile s
1 in c h e q u a ls 4 m ile s 1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 3 m ile s 1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 2 m ile s 1 in c h e q u a ls 1 .5 m ile s
1 in c h e q u a ls 1 .5 m ile
1 in c h e q u a ls 3 m ile s
1 in c h e q u a ls 2 m ile s
1 in c h e q u a ls 2 m ile s
( b ) 2 0 1 9 B u r g la r y H o t s p o t s
( c ) 2 0 1 8 , 2 0 1 9 a n d 2 0 2 0 c a t e g o r y 2 h o t s p o t a lo n g
w e s t e r n I - 6 1 0 .
( d ) 2 0 1 8 , 2 0 1 9 a n d 2 0 2 0 c a t e g o r y 2 h o t s p o t a t
C a n a l a n d 9 0 .
± ±
± ±
± ±
( e ) - ( g ) 2 0 1 8 - 2 0 2 0 B u r g la r y H o t s p o t s in ju s t 1 y e a r
±
± ±
± ±
± ±
±
2 0 1 8 B u r g l a r y
C r i m e H o t S p o t
L o w
M e d i u m
H i g h
V e r y H i g h
0 3 5 7 0 1 7 . 5 M i l e s
±
9 0
C a n a l
I - 4 5
S c o t t
I - 6 1 0
9 0
I - 6 9
9 0
S t e lla L in k
2 0 1 9 B u r g l a r y
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
H a r r i s C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
W e s t h e i m e r
K i r b y
3 3
2 0 2 0 B u r g l a r y
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
34
A large category 2 cluster occurred south of 288 and I-45 and northeast of 288 and I-69 in
all years. To the southwest of 288 and 45, there was a category 3 cluster at Sampson and Elgin in
10H50 in all three years. From 2019 to 2020, this category 3 hotspot gradually increased in size.
Just east of I-610 and 90, a category 3 cluster occurred on I-610 surrounded by a category
2 hotspot in 2018. In 2018 and 2019, the category 3 increased in size from 2018 to 2019 and then
the category 3 disappeared in 2020 and transformed into a medium sized hotspot. This category
2 hotspot occurred in all years along 90 near I-610 and was similar in size across all three years.
There was a medium category 2 hotspot between Old Spanish and I-610 along the 90 and
Main in 2018 and two small category 2 clusters south of I-610 in 2018. In 2019, the medium
sized cluster increased to a medium sized category 2 cluster between the 90 and I-610. The
small category 2 clusters south of I-610 in 2018 dramatically increased in size to larger category
2 clusters in 2019. In 2020, the medium sized category 2 cluster was still prevalent between the
90 and I-610. The large category 2 south of the I-610 morphed into three smaller category 2
clusters. In all years, a category 2 hotspot can be found at Wayside and Canal in the eastern
section of the study area.
In 2018 and 2019, a small category 2 occurred at Fulton and Morris, but it disappeared in
2020. There were three category 2 clusters in the northeast section in 2018 and 2019, but there
are just two larger category 2 hotspots in this area at Lockwood and Collingsworth and I-10 and
Gregg in 2020.
4.1.3 2018-2020 Summary of Robbery Hotspots

The robberies in 2018 occurred mainly in the midtown 10H40 and downtown 1A10
districts (Figure 7). A large category 5 hotspot occurred between the border of the downtown and
midtown districts on I-45. This was the same location as the 2019 and 2020 category 5 and 4
hotspots. An even larger category 4 and 3 hotspot surrounded the category 5 hotspot in the
0 3 0 6 0 1 5 M ile s
±
H a r r i s C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
( a ) R o b b e r y H o t s p o t s i n 2 0 1 8 ( b ) R o b b e r y H o t s p o t s i n 2 0 1 9
F i g u r e 7 . 2 0 1 8 - 2 0 2 0 R o b b e r y C r i m e H o t s p o t s
3 5
( c ) R o b b e r y H o t s p o t s i n 2 0 2 0
2 0 1 8 R O B B E R Y
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
2 0 1 9 R O B B E R Y
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
2 0 2 0 R O B B E R Y
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
4 5
6 1 0
6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
4 5
6 1 0 6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
4 5
6 1 0
6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 4 m ile s
36
downtown and midtown districts in 2019. This large category 4 and category 3 hotspot extended
from the central midtown to the western central downtown in 2019. A large category 5 and 4
hotspot occurred within this category 3 cluster in 2019. The remaining areas of the downtown
and midtown districts experienced category 2 hotspots. A small portion of the category 2 hotspot
spread into the eastern Montrose District. These hotspots were similar in size to the 2020
hotspots. A category 2 cluster occurred at Anita and Sampson in 10H50 in 2018. In 2019, this
area saw a large category 2 cluster that extended into the downtown and midtown districts. In
2020, this category 2 cluster reverted to a medium sized cluster at Elgin and Sampson.
It is surprising there are no hotspots showing up along Washington Avenue in 2018.

A small category 2 cluster occurred at 90/Main and Old Spanish in 15E40 in 2018. In 2019, this
category 2 became a medium sized category 2 cluster with a category 3 hotspot inside it. In
2020, this pair of hotspots disappeared. There was a small category 2 cluster north of I-610 and
west of 288 in 15E40 in 2018 and 2020.
A small category 2 cluster occurred at Old Spanish and La Sallette in 14D10 and 10H60
east of 288 in 2018. This cluster was not present in 2019, but it grew in 2020 to a medium sized
category 2 cluster along Scott with a northern border along the 90 and a southern border along I-
610. Northwest of the intersection of I-45 and 610, a large category 2 hotspot occurred between
Wayside and I-610 in 11H10, 11H20, and 13D10 in 2018. In 2019, this category 2 cluster
shrank and included a category 3 hotspot inside it at the intersection of 90 and I-45 and in 2020,
it became a medium sized category 2 cluster.
Figure 7 summarizes the 2018-2020 robbery trends and Figure 8 shows the robbery
hotspots that occurred in just one year, in two of the three years and those that occurred in all
three years. The no hotspot areas were not mapped for robbery.
F u lt o n
I - 6 1 0
9 0
9 0
I - 6 1 0
9 0
I - 6 1 0
I - 4 5
I - 4 5
I - 1 0
G r e g g
L ib e r t y
C a v a lc a d e
S c o t t
I - 6 9
M o r r is
A lm e d a
S c o t t
M L K
B e llf o r t
S u n b e a m
C u lle n
M L K
C a n a l
B e llf o r t
C u lle n
I - 6 1 0
9 0
I - 4 5
( a ) 2 0 1 8 a n d 2 0 2 0 R o b b e r y H o t s p o t s
( b ) 2 0 1 9 R o b b e r y H o t s p o t s
( f) 2 0 1 9 a n d 2 0 2 0 R o b b e r y H o t s p o t s in t h e S E s e c t io n . ( g ) 2 0 1 8 - 2 0 2 0 c a t e g o r y 2 h o t s p o t s n o r t h o f I - 1 0 a n d e a s t o f I - 6 9 .
( h ) 2 0 1 8 - 2 0 2 0 c a t e g o r y 2 h o t s p o t s in t h e s o u t h e r n s e c t io n .
( d ) 2 0 1 9 c a t e g o r y 3 a n d c a t e g o r y 2 h o t s p o t a t K ir b y a n d I - 6 9 .
T h e r e is a n o t h e r c a t e g o r y 2 h o t s p o t a lo n g w e s t e r n I - 6 1 0 .
( e ) 2 0 1 9 c a t e g o r y 2 b u r g la r y h o t s p o t a t C a n a l a n d 9 0 .
( i) 2 0 1 8 - 2 0 2 0 c a t e g o r y 3 a n d c a t e g o r y 2 h o t s p o t c lu s t e r
a t t h e in t e r s e c t io n o f 9 0 a n d I - 6 1 0 .
( j) 2 0 1 8 - 2 0 2 0 c a t e g o r y 2 h o t s p o t s in t h e s o u t h e r n s e c t io n .
( k ) d is p la y s t h e 2 0 1 8 a n d 2 0 1 9 R o b b e r y H o t s p o t s a t F u lt o n a n d M o r r is .
I - 6 1 0
9 0
9 0
I - 6 1 0
9 0
I - 6 1 0
I - 4 5
I - 4 5
I - 1 0
G r e g g
L ib e r t y
C a v a lc a d e
S c o t t
S u n b e a m
I - 6 1 0
I - 4 5
( a ) a n d ( b ) 2 0 1 8 - 2 0 2 0 R o b b e r y H o t s p o t s in a ll 3 y e a r s
( c ) - ( e ) 2 0 1 9 R o b b e r y H o t s p o t s
F ig u r e 8 . F r e q u e n c y o f R o b b e r y C r im e H o t s p o t s
K ir b y
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 2 .5 m ile s 1 in c h e q u a ls 3 m ile s
1 in c h e q u a ls 2 m ile s
1 in c h e q u a ls 2 .5 m ile s
1 in c h e q u a ls 2 m ile s
1 in c h e q u a ls 1 .5 m ile s
1 in c h e q u a ls 2 m ile s
1 in c h e q u a ls 3 m ile s
1 in c h e q u a ls 2 m ile s
2 0 2 0 R o b b e r y
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
2 0 1 9 R o b b e r y
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
2 0 1 8 R o b b e r y
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
b .
H a r r is C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
0 3 0 6 0 1 5 M i l e s
( i. ) - ( k . ) R o b b e r y H o t s p o t s in 2 y e a r s
( f) - ( h ) R o b b e r y H o t s p o t s in a ll 3 y e a r s
W a s h in g t o n A v e
±
±
±
±
±
± ±
±
±
± ±
±
L o c k w o o d
I - 6 9
3 7
( c ) 2 0 1 9 c a t e g o r y 3 a n d 2 c lu s t e r a lo n g W a s h in g t o n A v e .
38
Another small category 2 cluster exists east of I-69 in 7C20 in 2018 between Roland and
Liberty. To the east of the I-45, a category 2 cluster occurred at Fulton and Luzon in 2A10 in
2018. In 2019, a medium sized category 2 occurred at Fulton and Morris and a large category 2
hotspot occurred at Cavalcade and Lockwood. There was no category 2 hotspots at Fulton and
Morris in 2020. There were 2 medium sized category 2 clusters at Lockwood and the I-610 and
the I-10 and Lockwood in 2020. There was a small category 2 cluster at Collingsworth and the I-
69 in 2020. To the east of the I-45 and north of the I-10 in 2A10, there is a medium sized
category 2 cluster on Fulton Street in 2019.
In the southern section of the study area, an extremely small category 2 cluster occurred
west of Cullen, north of the I-610 and south of the 90 in 2019. A small category 2 hotspot
occurred along Cullen south of the 90 and north of the I-610. Another small category 2 cluster
spanned the 90 west of MLK. South of the I-610, three small and one medium sized category 2
exist. At Scott and Bellfort, a small category 2 occurred. A medium sized category 2 occurred
near MLK and Bellfort. A medium sized category 2 also occurred along Telephone near the I-
610. A small category 2 occurred near Bellfort and Cullen and another small category 2 hotspot
also occurred at Scott and Wilmington as well.
In the southeastern section, another category 2 hotpot occurred at the I-610 and
Telephone in 13D10 in 2020. Category 2 hotspot clusters occurred at Wayside and the I-45,
Almeda and the I-610, Sunbeam and Cullen, Scott between Bellfort and Sunbeam, Bellfort and
MLK, and at Telephone and the I-610 in 2020. In the western section of the study, a medium
sized category 2 hotspot occurred near Westheimer and the I-610 in 2019.
A large category 5 robbery cluster occurred at Milam and the I-45 in 2019. This category 5
is very close to the category 5 locations in 2018 and 2020 in the northwestern midtown district. A
very large category 4 cluster surrounded this category 5 cluster in 2019. It occurred in the
39
northwestern midtown district, southwestern to central downtown district, and the eastern
Montrose district in 2019.
The central to eastern Montrose district had a large category 4 hotspot in 2019, close to
the same location as the category 2 hotspot in 2020. The majority and remaining areas of the
Montrose district 1A20, downtown 1A10 district, and midtown 10H40 district saw large
category 3 and category 2 robbery hotspots in 2019. There was a category 3 cluster at Lake and
I-69 near the Kirby District in 1A30. This category 2 spread from the downtown and midtown
districts all the way to the western Upper Kirby District in 2020.
Along Washington Avenue at Durham, a large category 3 cluster occurred in 2019. There
was a very large category 2 cluster that extended from Washington Avenue and the I-10 at
Westcott to Sawyer in the east. This category 2 connects to the Montrose district hotspots to the
south on Studemont. There were no robbery hotspots along Washington Avenue in 2018 and
2020.
There were category 5 and 4 robbery hotspots near the intersection of downtown 1A10
and midtown 10H40 districts in 2020 at the intersection of the I-45 and Brazos. A large circular
category 5, 4, and 3 hotspot occurred in the northwest portion of the midtown district near the I-
45. This category 5 hotspot location is the same as the 2019 and 2018 category 5 robbery
hotspots. The remainder of the midtown and downtown districts saw category 3 and 2 hotspots
in the southern portion of the district just like in 2018. In 2019, these category 3 and 2 hotspots
intensified into large category 4 and 3 hotspots in the midtown and downtown districts. There
was no occurrence of robbery hotspots along Washington Ave in the bar district in 2020, most
likely due to COVID-19 and the bars being shut down.
40
There were category 2 robbery clusters spread throughout the study area in 2020. In the
northwestern section of the study area, a medium sized category 2 occurred at 19
th
and Ashland.
In the southern portion of the study area, a large category 2 cluster occurred east of 288 and
north of the I-610. This category 2 cluster was located in police beats 14D20 and 14D10 and
spread from the west to the east along the I-610 for nearly two miles.
4.1.4 2018-2020 Summary of Theft Hotspots

There was a large category 5, 4 and 3 theft hotspot that intersected the downtown 1A10,
midtown 10H40, and Montrose 1A20 police districts near the I-45 in 2020 (Figure 9). The
category 5 hotspot was located at the intersection of Baldwin and Gray on the northwestern
border of the midtown district. The midtown district experienced all three category 5 hotspots. In
the northwestern midtown district, a category 5 to 4 hotspot occurred. A category 2 hotspot is
found in the southwestern corner of this district as well. The downtown district also experienced
category 3 and 2 hotspots that extended from the southwest to the central portion of the district.
A category 4 and 3 hotspot occurred at Kirby and the I-69 in 1A30. A large category 2 hotspot
connected the Montrose, downtown and midtown hotspots as well. There were two small
category 3 clusters near Washington Avenue in 2020. A large category 2 hotspot extended from
Washington Ave to the Montrose, downtown and midtown high theft hotspots.
A large category 2 hotspot occurred south of the I-69 and east of 288 in 2019. This
cluster was located in the 10H60 and 10H50 police districts. A small category 3 cluster inside
the large category 2 cluster was located near the intersection of Elgin and Sampson as well.
These clusters disappeared in 2020.
Large category 5 and 4 theft hotspots occurred in the northwestern portion of the
midtown district at Milam and the I-45 in 2019 near the same category 5 and 4 hotspot that
occurred in 2020. This hotspot spanned the Montrose 1A20, downtown 1A10, and midtown
0 3 0 6 0 1 5 M ile s
±
H a r r i s C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
( a ) T h e f t H o t s p o t s i n 2 0 1 8 ( b ) T h e f t H o t s p o t s i n 2 0 1 9
F i g u r e 9 . 2 0 1 8 - 2 0 2 0 T h e f t C r i m e H o t s p o t s
4 1
( c ) T h e f t H o t s p o t s i n 2 0 2 0
2 0 1 8 T H E F T
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
2 0 1 9 T H E F T
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
2 0 2 0 T H E F T
C R I M E H O T S P O T S
L O W
M E D I U M
H I G H
V E R Y H I G H
4 5
6 1 0
6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0 9 0
4 5
6 1 0
6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
4 5
6 1 0 6 9
1 0
6 9
9 0
6 1 0
6 1 0
2 8 8
4 5
6 1 0
9 0
42
districts. In 2019, category 4 and 5 hotspots emerged in the central Montrose District. The two
category 5 hotspots were connected by a narrow category 4 hotspot in 1A20 in 2019 that did not
occur in 2020. The central category 4 and 5 hotspot in Montrose spread out into a category 3
hotspot in 2019 that was more expansive than the category 3 hotspot in 2020 (Figure 9). The
southwest corner of the downtown district contained a category 4 hotspot as well.
The remaining area of the Montrose district was a category 2 theft hotspot. Both 2020
and 2019 had category 2 clusters that covered most of the downtown, midtown, and Montrose
districts and some of the Upper Kirby District. The 2019 hotspot was larger and extended
southeast to the intersection of the I-45 and 288.
Category 4 and 3 clusters were found at the Upper Kirby district and the I-69 in both 2019
and 2020. Both clusters were larger in 2020. The remaining portion of the Kirby District was a
category 2 cluster that expanded into the Montrose District in both years. The category 2 cluster
was larger in 2019. Just north of Washington Avenue in the bar district, a category 4 hotspot
occurred at the intersection of Durham and Allen in the 2A50 police precinct in 2019. A
category 3 hotspot extended from Reinerman to the west to Patterson to the east along
Washington Ave. A large category 2 hotspot also extended west to east along Washington Ave
from TC Jester to Heights.
In 2020, the Washington Ave bar district had two category 3 hotspots, connected by a
large category 2 cluster in police precinct 2A50, that connects to the downtown and midtown
districts at Studemont. There was also a category 3 cluster at Allen and Rochow south of
Washington Ave, and another to the northeast of the I-45 and the I-69. In 2019, this category 3
extended into the downtown district and was connected to the large category 5, 4, and 3 hotspots
in the downtown and midtown districts (Figure 9). These category 5, 4, and 3 hotspots in the
43
Montrose district in 2019 diminished to small category 3 and large category 2 hotspots in 2020.
In 2019, there was one category 4 and one category 3 cluster along Washington Avenue
that were surrounded by a category 2 cluster. This category 2 did not connect with the clusters in
the midtown and downtown districts in 2019. There was also two small category 2 clusters near
Westheimer and the I-610 in 2019 that morphed into category 3 clusters surrounded by category
2 clusters in 2020. In 2019, category 3 and category 2 hotspots occurred at the intersection of I-
610 and Main/90, but the category 3 hotspot disappeared and the category 2 hotspot diminished
in size in 2020.
The northeastern section of the study area had five small category 2 hotspots and one
medium sized category 2 hotspot in 2019. Three of the category 2 hotspots are located on the I-
10, near Wayside, Gregg and Lockwood. The final two small category 2 hotspots occurred at
Gregg and Liberty, and the I-610 and Lockwood. The medium sized category 2 hotspots are
located at Cavalcade and Lockwood. None of these category 2 hotspots occurred in 2020.
There are three clusters of category 2 theft hotspots in the southern portion of the study
area, located north of the I-610 and south of Old Spanish between the 288 and MLK in police
precincts 14D10 and 14D20. The largest category 2 hotspot spreads north and south along MLK
between Old Spanish and the I-610. None of these category 2 hotspots were present in 2020
except for a very small category 2 hotspot near Calhoun and MLK. There were two small
category 2 hotspots south of I-610, one at Sunbeam and Scott and the other located west of Cullen
and north of Sunbeam in 2019. South of I-610 at Bellfort and MLK, there was also a category 2
theft hotspot located in police precinct 14D30. This hotspot extends north and south along MLK.
None of these theft hotspots occurred in 2020. Figure 10 shows the 2019 and 2020 theft hotspots
that occurred in both years, no years and the hotspots that occurred in just one year.
F ig u r e 1 0 . F r e q u e n c y o f T h e f t C r im e H o t s p o t s
I - 6 9
( a ) 2 0 1 9 a n d 2 0 2 0 T h e f t H o t s p o t s
( b ) 2 0 1 9 T h e f t H o t s p o t s
( h ) 2 0 1 9 c a t e g o r y 4 , c a t e g o r y 3 a n d c a t e g o r y 2 h o t s p o t c lu s t e r a lo n g W a s h in g t o n
A v e . T h e r e is a s m a ll 2 0 2 0 c a t e g o r y 4 , 3 a n d c a t e g o r y 2 in t h e s a m e lo c a t io n .
T h e r e a r e t w o a d d it io n a l c a t e g o r y 3 c lu s t e r s in 2 0 2 0 .
( g ) 2 0 2 0 c a t e g o r y 4 , 3 , a n d 2 h o t s p o t c lu s t e r a t I - 6 9 a n d K ir b y . T h e r e is a 2 0 1 9 c a t e g o r y 3 a n d c a t e g o r y 2
in t h e s a m e lo c a t io n . T h e r e a r e t w o 2 0 2 0 c a t e g o r y 3 a n d c a t e g o r y 2 h o t s p o t c lu s t e r s n e a r I - 6 9 a n d I - 6 1 0 .
A 2 0 1 9 c a t e g o r y 2 o c c u r s in t h e s a m e lo c a t io n .
( j) C a t e g o r y 3 a n d C a t e g o r y 2 c lu s t e r a t 9 0 a n d I - 6 1 0 .
( c ) 2 0 1 9 c a t e g o r y 2 h o t s p o t s i n t h e s o u t h e r n s e c t i o n . ( d ) 2 0 1 9 c a t e g o r y 2 h o t s p o t s i n t h e S E s e c t i o n .
( f ) 2 0 1 9 c a t e g o r y 2 t h e f t h o t s p o t a t C a n a l a n d 9 0 . ( e ) 2 0 1 9 c a t e g o r y 2 h o t s p o t s n e a r t h e i n t e r s e c t i o n
o f L o c k w o o d a n d C o l l i n g s w o r t h .
( i) 2 0 1 9 a n d 2 0 2 0 c a t e g o r y 2 h o t s p o t s in t h e N o r t h w e s t s e c t io n .
I - 6 1 0
9 0
9 0
I - 6 1 0
I - 6 1 0
9 0
I - 4 5
I - 1 0
I - 6 1 0
C a n a l
W a s h i n g t o n A v e
W e s t h e i m e r
A l m e d a
A s h l a n d
1 9 t h
B e v i s
1 in c h e q u a ls 3 m ile s
1 in c h e q u a ls 2 m ile s
1 in c h e q u a ls 2 m ile s 1 in c h e q u a ls 3 m ile s
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 4 m ile s 1 in c h e q u a ls 4 m ile s
1 in c h e q u a ls 3 m ile s 1 in c h e q u a ls 2 m ile s
±
±
( a ) a n d ( b ) 2 0 1 9 - 2 0 2 0 T h e f t H o t s p o t s i n 2 y e a r s
( c ) - ( f) 2 0 1 9 T h e f t H o t s p o t s in ju s t o n e y e a r
( g ) - ( j) 2 0 1 9 - 2 0 2 0 T h e f t H o t s p o t s in 2 y e a r s
( k ) a n d ( l) A r e a s w it h n o t h e f t h o t s p o t s
2 0 2 0 T h e f t
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
± ±
± ±
±
± ±
± ±
±
S c o t t
M L K
Y e llo w s t o n e
G r ig g s
B e llf o r t
S u n b e a m
C u lle n
M L K
B e llf o r t C u lle n
S u n b e a m
S c o t t
I - 4 5
I - 6 9
I - 1 0
G r e g g
L ib e r t y
C a v a lc a d e
L o c k w o o d
I - 6 9
I - 6 1 0
C o llin g s w o r t h
H a r r i s b u r g
I - 6 1 0
I - 6 9
I - 6 1 0
9 0
H a r r i s C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
±
( k ) S W s e c t io n o f t h e s t u d y w it h n o h o t s p o t s . ( l) S E s e c t io n o f t h e s t u d y w it h n o h o t s p o t s .
1 in c h e q u a ls 3 m ile s 1 in c h e q u a ls 2 m ile s
2 0 1 9 T h e f t
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
K ir b y
4 4
45

4.1.5 Crime Types by Year
Figures 11-13 show all the crime types by year. Figure 11 shows the 2018 assault,
burglary, robbery and theft crimes. Figure 12 shows the 2019 assault, burglary, robbery and theft
crimes. Figure 13 shows the 2020 assault, burglary, robbery and theft crimes. In 2020, all crime
types saw category 5 and 4 hotspots on the I-45 near the downtown and midtown districts. The
same area experienced the category 5 and 4 clusters in 2019, but this cluster expanded into the
eastern Montrose districts as well. In 2018, it was confined to a smaller area than in 2020.
4.2 2020 Land Use Types for Optimized Crime Hotspots
The 2020 optimized hotspots were determined for the crime types as a test of the veracity
of these crime data. This test compared hotspots using two different ArcGIS data tools, Kernel
Density Analysis and Optimized Hotspot Analysis. The Land Use Layer was downloaded from
the COHGIS. Once the Crime Hotspots were determined, the intersect tool was used to
determine the Land Use attributes of the crime hotspots areas for each crime type in 2020. Land
uses, such as bars (commercial land use), were hypothesized to have contributed to crime
hotspots.
The 2020 Assault Optimized Hotspots occurred in the midtown and downtown districts,
with Residential: Single-Family (48%) and Multi-Family (18%) and Commercial (4%). Another
crime hotspot occurred in the southern portion of the study area where Residential: Single-
Family made up 68% and undeveloped land covered 24% of this area. A similar crime hotspot in
the northeastern portion of the study area was made up of Residential: Single-Family (61%) and
Undeveloped Land (33%) as well.
The 2020 Burglary Optimized Crime Hotspot occurred in the midtown, downtown,
Montrose and part of the Upper Kirby district, consisting of Residential: Single-Family (55%),
Multi-Family (17%), Undeveloped land (17%), Commercial (5%), and Industrial (3%). At US 90
( a ) 2 0 1 8 A s s a u l t H o t s p o t s
F i g u r e 1 1 . 2 0 1 8 C r i m e s b y T y p e
0 3 0 6 0 1 5 M ile s
±
H a r r i s C o u n t y , T X S t u d y A r e a ( O r a n g e )
2 0 1 8 A s s a u lt
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
2 0 1 8 B u r g la r y
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
2 0 1 8 R o b b e r y
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
2 0 1 8 T h e f t
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
1 0
( b ) 2 0 1 8 B u r g l a r y H o t s p o t s
( c ) 2 0 1 8 R o b b e r y H o t s p o t s
( d ) 2 0 1 8 T h e f t H o t s p o t s
1 i n c h e q u a l s 5 m i l e s 1 i n c h e q u a l s 5 m i l e s
1 i n c h e q u a l s 5 m i l e s
1 i n c h e q u a l s 5 m i l e s
4 6
( a ) 2 0 1 9 A s s a u l t H o t s p o t s
F i g u r e 1 2 . 2 0 1 9 C r i m e s b y T y p e
0 3 0 6 0 1 5 M ile s
±
H a r r i s C o u n t y , T X S t u d y A r e a ( O r a n g e )
2 0 1 9 A s s a u lt
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
2 0 1 9 B u r g la r y
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
2 0 1 9 R o b b e r y
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
2 0 1 9 T h e f t
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
1 0
( b ) 2 0 1 9 B u r g l a r y H o t s p o t s
( c ) 2 0 1 9 R o b b e r y H o t s p o t s ( d ) 2 0 1 9 T h e f t H o t s p o t s
1 i n c h e q u a l s 5 m i l e s 1 i n c h e q u a l s 5 m i l e s
1 i n c h e q u a l s 5 m i l e s
1 i n c h e q u a l s 5 m i l e s
4 7
( a ) 2 0 2 0 A s s a u l t H o t s p o t s
F i g u r e 1 3 . 2 0 2 0 C r i m e s b y T y p e
0 3 0 6 0 1 5 M ile s
±
H a r r i s C o u n t y , T X S t u d y A r e a ( O r a n g e )
2 0 2 0 A s s a u lt s
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
2 0 2 0 B u r g la r y
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
2 0 2 0 R o b b e r y
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
2 0 2 0 T h e f t
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
1 0
( b ) 2 0 2 0 B u r g l a r y H o t s p o t s
( c ) 2 0 2 0 R o b b e r y H o t s p o t s ( d ) 2 0 2 0 T h e f t H o t s p o t s
1 i n c h e q u a l s 5 m i l e s 1 i n c h e q u a l s 5 m i l e s
1 i n c h e q u a l s 5 m i l e s
1 i n c h e q u a l s 5 m i l e s
4 8
49

and the I-610, another crime hotspot occurred Single-Family (75%), Commercial (11%), and
Undeveloped land (6%). The 2020 Robbery Optimized Crime Hotspots occurred in the
downtown and midtown districts, with predominately Residential: Single-Family (47%), Multi-
Family (29%), and Undeveloped land (20%).
The 2020 Theft Optimized Land use occurred in the downtown, midtown, Montrose, and
Kirby districts. This large crime hotspot extended into the bar district along Washington Avenue.
The land use types for this crime hotspot included Residential: Single-Family (55%), Multi-
Family (17%), Commercial (5%), and Industrial (3%). The second theft crime hotspot occurred
in the western portion of the Kirby District. The land use types for this crime hotspot included
Residential: Single-Family (67%) and Multi-Family (25%) and another theft hotspot located
along I-610 in the western section of the study comprised of Multi-Family (54%), and Single-
Family (38%) residential land uses.
Figure 14 shows the 2020 optimized hotspots for assaults, burglaries, robberies and theft.
All four crime types show optimized hotspots in the downtown and midtown districts in 2020.
Figure 15 shows land use in the downtown and midtown districts. Ten land use categories occur
in this part of the study area.
The 2020 Assault Optimized Hotspots occurred mainly in the downtown, midtown and
Montrose districts. The northeastern section saw an area of Gi Bin level 3 assault and the
southern section of the study saw a large Gi Bin level 3 assault south of US 90, east of 288 and
south of the I-610. 2020 Burglary was confined to the downtown, midtown, Montrose and Upper
Kirby Districts except for a very small area at US 90 near the Astrodome. 2020 robbery was
confined to just a small cluster at the downtown, midtown, and Montrose districts. 2020 theft
occurred in a large area generally south of the I-10 and north of the I-69 and east of the I-610.
The highest theft cluster was at the downtown, midtown, Montrose, Kirby and Washington Ave.
F i g u r e 1 4 . 2 0 2 0 O p t i m i z e d C r i m e H o t s p o t s
0 3 0 6 0 1 5 M ile s
±
B u r g la r y O p t im iz e d H o t s p o t s
R o b b e r y O p t im iz e d H o t s p o t s
T h e f t O p t im iz e d H o t s p o t s
A s s a u lt O p t im iz e d H o t s p o t s
H a r r i s C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
1 0
4 5 6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
1 0
4 5
6 9
6 9
9 0
6 1 0
6 1 0
( a ) 2 0 2 0 A s s a u l t O p t i m i z e d H o t s p o t s ( b ) 2 0 2 0 B u r g l a r y O p t i m i z e d H o t s p o t s
( c ) 2 0 2 0 R o b b e r y O p t i m i z e d H o t s p o t s ( d ) 2 0 2 0 T h e f t O p t i m i z e d H o t s p o t s
6 1 0
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
2 8 8
2 8 8
2 8 8
9 0
9 0
5 0
1 i n c h e q u a l s 5 m i l e s
± ±
±
±
1 i n c h e q u a l s 5 m i l e s
1 i n c h e q u a l s 5 m i l e s 1 i n c h e q u a l s 5 m i l e s
4 5
9 0
6 9
L a n d U s e ( G r o u p e d )
S in g le - F a m ily R e s id e n t ia l
M u lt i- F a m ily R e s id e n t ia l
C o m m e r c ia l
O f f ic e
P u b lic & I n s t it u t io n a l
I n d u s t r ia l
P a r k & O p e n S p a c e s
T r a n s p o r t a t io n & U t ilit y
U n d e v e lo p e d
A g r ic u lt u r e P r o d u c t io n
U n k n o w n
W a s h in g t o n A v e n u e
0 0 .5 1 0 .2 5 M ile s
F i g u r e 1 5 . D o w n t o w n a n d M i d t o w n D i s t r i c t L a n d U s e s 5 1
1 0
4 5
6 9
1 0
2 8 8
52

4.3 Crime Statistics Before and After Natural Disasters

One month after Tropical Storm Beta in September 2020, Assault decreased by 63
occurrences and Burglary decreased by 73 occurrences (Table 3). In September 2020, Robbery
increased by 18 occurrences and Theft increased by 23 occurrences. Assault in October 2020 saw
32 more occurrences than in August 2020 and 95 more occurrences than in September 2020.
Burglary in October 2020 had 29 less occurrences than in August 2020 and 44 more
occurrences in October 2020 than in September 2020 (Table 3). Robbery in October 2020
increased by 2 occurrences than in August 2020 and fell by 16 occurrences in October 2020
compared to September 2020. Theft in October 2020 saw 93 more occurrences than in
September 2020 and 116 more occurrences than in August 2020 (Table 3).
Table 3. Crime Statistics Before and After Tropical Storm Beta

Crime Type August 2020 % Change September
2020
% Change October 2020

Assault 1075 - 6 1012 + 9 1107
Burglary 415 - 18 342 + 13 386
Robbery 182

+ 10 200

- 8 184
Theft 798 + 3 821 + 11 914
Source: Houston Police Department

Tropical Storm Imelda occurred in September 2019. From August 2019 to September
2019, assault decreased by 8% (Table 4). October 2019 Assault saw a 3% decrease in crime
compared to August 2019. From August 2019 to September 2019, burglary decreased by 2%.
October 2019 Burglary saw a 11% decrease in crime type compared to August 2019. From
August 2019 to September 2019, Robbery decreased by 9%. In October 2019, Robbery also
decreased by 9% compared to August 2019. Theft in September 2019 decreased by 10%
compared to August 2019. Theft in October 2019 decreased by around 11% compared to August
2019 (Table 4).
53

Table 4. Crime Statistics Before and After Tropical Storm Imelda

Source: Houston Police Department

Hurricane Harvey made landfall on the Texas coast on August 27
th
, 2017 and the percent
changes for crime before and after are shown in Table 5. The increases and decreases of crimes
in specific months are compared to July 2017. For example, there were 24 less assault crimes in
August 2017 compared to July 2017.
Table 5. Crime Statistics Before and After Hurricane Harvey

Crime
Type
July
2017
% Change August
2017
%
Change
September
2017
%
Change
October
2017

Assault 375 - 6 351 - 3 341 + 8 369
Burglary 419 + 35 564 - 28 408 - 5 388
Robbery 231 - 13 200 - 19 162 + 47 238
Theft 2226 - 10 2013 - 13 1754 + 16 2027
Source: Houston Police Department

4.4 Bus Hotspot Locations and 2020 Crime Hotspot Locations

There are 4,106 Metro Bus Stops within the I-610 Loop study area and just less than 10%
(307) of these bus stops fall within the downtown and midtown districts that contain the police
precincts 1A10 and 10H40. Kernel Density analysis on the bus stop locations showed category
5, 4, 3, and 2 densities of bus stops within the downtown and midtown districts. The downtown
district has a large category 5, 4, 3 covering most of the central portion of the downtown district.
The category 5, 4 and 3 hotspots falls in the downtown 1A10, midtown 10H40, and 1A20
Montrose districts. Appendix C displays the kernel density analysis on the Metro Bus Stops.
Crime Type August 2019 % Change September
2019
% Change October 2019

Assault 1112 - 8 1021 - 5 1076
Burglary 477 - 3 464 - 8 423
Robbery 252 - 10 228 0 228
Theft 1499 - 10 1342 - .02 1338
54

The northern to central portion of the midtown district has a category 4 and 3 density of
bus stops. The southern portion of the midtown district has a category 2 density of bus stops. The
downtown and midtown districts are the only places in the study area with category 5, 4, 3
densities of bus stops. The remainder of the study area has category 2 densities. These category
2 densities of bus stops occur within 3-5 miles of the category 5,4,3 densities located within the
downtown and midtown districts.
This study also found that 2018-2020 crime type category 5 and 4 hotspots mainly
occurred in the downtown and midtown districts. There is a connection between very high
(category 5) and high (category 4) crime type hotspots and very high (category 5) and high
(category 4) bus stop locations. Appendix A shows the category 5 and category 4 densities of
bus stops occurring in the downtown and midtown districts. This is the same area that
experiences the category 5 and category 4 hotspots of Assault, Burglary, Robbery and Theft.
4.5 2020 Edge Effects Test of Data: July-December 2020

4.5.1 2020 Assault Boundary Hot Spots

There was no category 5, 4, and 3 assault hotspots that occurred outside the I-610
boundary. A small category 2 cluster occurs north of the I-610 at the intersection of Cross
Timbers and the I-45 in police precincts 3B40 and 3B50. Another small category 2 cluster
occurred at Cross Timbers and the I-69 in police precinct 7C30. A very small category 2 cluster
occurred at Homestead and Tidwell in police precinct 8C10. Beyond the western boundary of the
I-610, a larger category 2 cluster occurred at Rampart and Gulfton in 17E10. At Westheimer and
the I-610, a small category 2 cluster can be found. South of the I-610, a large category 2 hotspot
occurred on Cullen Street in police precinct 14D20. Southeast of the I-610, a small category 2
cluster occurred at Rockhill and Broadway in police precinct 13D20 and nearby at Howard and
55
the I-45, a very small category 2 cluster can be found in police precincts 13D20 and 11H30.
4.5.2 2020 Burglary Boundary Hot Spots
North of the I-610 study area, there was a small category 2 cluster near the intersection of
Bennington and the I-69 in police precinct 7C30. Just north of this intersection, a smaller
category 2 cluster occurred at Aldine Westfield and Jensen in police precincts 7C30, 8C10, and
7C20.
A very large category 2 burglary hotspot occurred to the north and northwest of the study
area, between Wirt and Fulton in police precincts 3B50, 3B40, 3B30, 3B10, 5F10, 2A60, 2A30,
and 2A20. The southern boundary extends well south of I-610 all the way to Fulton. Northwest
of the I-610 study area, a small category 3 cluster exists within this massive category 2 cluster at
the intersection of 34
th
and Mangum in police precinct 3B10. North of this category 3 cluster,
another small category 3 cluster surrounded by a large category 2 cluster occurred at Antoine
and Tidwell in police precinct 3B10.
An extremely small category 2 cluster was found at Antoine and the I-10 west of the
study area in police precincts 5F10 and 18F10. Another large category 2 cluster extends from
Mobud and Braewick to the intersection of Briar Hollow E and Briar Hollow N in police
precincts 17E10, 18F20, 1A50 and 1A40. Just outside the I-610 west of the study area, a small
category 4 cluster was located at Sage and Alabama in police precinct 18F20. This category 4
was surrounded by a large category 3 cluster that extends from Yorktown and Schumacher to
Post Oak and Post Oak Park across the I-610.
A small category 2 cluster was found at Braeswood and Braesmont southwest of the
study area boundary in police precinct 15E10. South of the I-610 study area, two small category
2 clusters extend as far as Bellfort and Cullen in police precincts 14D20 and 14D30. From the
56

southeast, two category 2 clusters extend as far as Sunbeam in police precincts 14D20 and
14D30 and Winfree (13D10, 11H20, and 11H10). There is also a category 3 cluster surrounded
by a category 2 cluster located southeast of the I-610 study area. Similar to the other crime types,
there were no burglary hotspots located to the east of the I-610 study area.
4.5.3 2020 Robbery Boundary Hot Spots
Outside the northern I-610 boundary, there is a medium sized category 4 robbery hotspot
at the intersection of Cross Timbers and the I-45. This category 4 hotspot is surrounded by a
category 3 hotspot in police precincts 3B40 and 3B50 and a category 2 hotspot surrounds the
entire category 3 hotspot as well. There is also a large category 2 hotspot at I-69 and Keeland in
police precincts 3B40 and 3B50 and a very small category 2 cluster occurred at Tidwell and
Homestead in police precincts 7C30 and 8C10.
Northwest of the I-610 boundary, a very small category 3 cluster can be found at the 290
and Kingswood police precinct in 3B10. A large category 2 cluster also occurred in this police
precinct 3B10 as well. Southwest of the I-610 study area, a category 5 can be found at Rampart
and Clarewood in police precinct 17E10. This category 5 robbery cluster is surrounded by
category 4 and 3 hotspot clusters. There is also a category 2 cluster that extends northeast of the
category 5 and category 4 clusters and a small category 3 cluster occurred inside this category 2
at the intersection of Mccue and Alabama in the Galleria area of police precinct 18F20.
A category 2 robbery hotspot extended south of the I-610 between Scott and Jutland to
Airport in police precinct 14D20. A small category 3 cluster surrounded by a category 2 cluster
can be found at Bellfort and Glencrest in police precincts 13D20 and 11H30. A medium sized
category 2 hotspot can be found at Laura Koppe and the I-69 to the north of the I-610. This
category 2 cluster extends north to Parker and south to Bennington along the I-69. Finally, a
57

small category 2 hotspot can be found at the intersection of Tidwell and Homestead in police
precincts 7C30 and 8C10.
4.5.4 2020 Theft Boundary Hot Spots
North of the I-610 study area, a very small theft cluster can be found at I-45 and Cross
Timbers in police precincts 3B40 and 3B50. Northwest of the I-610 study area, a very small
category 2 cluster can be found at 290 and 34
th
street in police precinct 3B10. West of the study
area near I-610, small category 5 and category 4 theft hotspots can be found at the intersection of
Westheimer and the I-610. Medium sized category 3 and category 2 theft hotspots surround the
category 5 and 4 hotspot in police precincts 18F20, 17E10, and 1A50.
A small category 2 cluster can also be found at the intersection of Hillcroft and
Clarewood in police precinct 17E10. Finally, another small category 2 cluster extends south of
the I-610 study area to Bellfort near US 90/Main in police precinct 15E40.
Figure 16 shows the 2020 edge effect crimes of assault, burglary, robbery and theft.
Appendix B shows the top 50 bars locations in Houston, TX. These data was obtained from
Google maps. The majority of these bars are located along Washington Ave in the bar district.
Several others occur in the downtown and midtown districts. There is a small cluster of these bar
locations near the I-45 in close proximity to the downtown and midtown districts. This area saw
the category 5 and 4 hotspot clusters for all crime types. There is also a small cluster of bars in
the Upper Kirby district near the intersection of the I-69 and Kirby Drive and this area saw a
small category 3 hotspot cluster for most of the crime types as well.
The final chapter that follows reviews the successes, challenges and failures of this study
and provides a discussion of the implications of the study for law enforcement and future
research on crime.
F i g u r e 1 6 . J u l y - D e c e m b e r 2 0 2 0 E d g e E f f e c t s
0 3 0 6 0 1 5 M ile s
±
H a r r i s C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
B u r g l a r y
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
T h e f t
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
A s s a u l t
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
( a ) 2 0 2 0 A s s a u l t E d g e E f f e c t s
( b ) 2 0 2 0 B u r g l a r y E d g e E f f e c t s
( c ) 2 0 2 0 R o b b e r y E d g e E f f e c t s ( d ) 2 0 2 0 T h e f t E d g e E f f e c t s
A s s a u l t
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
1 0
4 5
6 9
4 5
9 0
6 1 0
6 1 0
4 5
R o b b e r y
C r i m e H o t S p o t s
L o w
M e d i u m
H i g h
V e r y H i g h
6 9
4 5
9 0
6 1 0
6 1 0
4 5
6 1 0
1 0
4 5
6 9
6 9
4 5
1 0
4 5
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
1 0
4 5
6 9
6 9
9 0
6 1 0
6 1 0
6 1 0
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
1 0
4 5
6 9
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
9 0
2 8 8
2 8 8
2 8 8
2 8 8
9 0
± ±
± ±
1 i n c h e q u a l s 6 m i l e s 1 i n c h e q u a l s 6 m i l e s
1 i n c h e q u a l s 6 m i l e s
1 i n c h e q u a l s 6 m i l e s
5 8
59

Chapter 5 Conclusions

Houston, Texas has one of the highest rates of crime in the nation. To address this threat, law
enforcement must understand the spatial and temporal patterns of crime, as well as what
variables are related to criminal activity. This project sought to visualize the distribution of crime
types in downtown Houston between 2018 and 2020 to enable law enforcement agencies to
better respond to disturbances in the region. While the area was bounded by the inner I-610 loop,
edge effects were taken into consideration. The particular crime types analyzed were assault,
burglary, theft, and robbery hotspots in the inner I-610 Houston, Texas region. Additional
variables included in the study, which were identified through the literature review, included
land use type and proximity to bus stations. An exploratory component of the project looked at
crime rates before and after natural disasters, namely a tropical storm or hurricane.
Crime location data was supplied by the Houston Police Department Crime Statistics
website. The data on land use type and bus station location were provided by the City of Houston
GIS. Once all the data was cleaned, aggregated, and uploaded to ArcGIS Pro, kernel density
analysis and hot spot analysis were used to visualize crime hotspots. The crime addresses were
geocoded using CDX WinZip and ArcGIS Pro Geocoder to facilitate further spatial analysis and
visualization. A number of visualizations were created to show the target audience how crime
type and location changed over time. Analysis was conducted looking at the relationship between
crime, bus stop locations, and land use types.
The midtown, downtown and Montrose districts experienced the most category 5, 4 and 3
crime hotspots for the four crime types for all three years of the study. The highest density of bus
stops were located within the downtown and midtown districts of the study area. The land use
type that represented the largest areas (in acres) for the optimized hotspots was mainly single-
60

family: residential. Although the largest land use type within these Gi Bin +3 crime hotspots was
single-family, this does not necessary mean that single-family units are the cause of the crime
hotspots.
Generally, crime was more prevalent and more widespread in 2019 than in 2018 and
2020. Crime trends showed category 5 and 4 hotspots spreading into the Montrose district and
category 3 and category 2 hotspots expanding to the Washington Avenue bar district in 2019. For
some of the crime types, this expansion was not evident in 2018 and 2020. Crime may not have
been as prevalent in 2020 because of the closures, curfews and Covid pandemic lockdown of
2020. This could possibly explain, for example, why assault was present along Washington
Avenue in 2019 and not in 2020.
5.1 Research Questions and Summary of Findings
This study addressed the following research questions in the inner I-610 Houston, TX:
1) Where are crime hot spots located within the city of Houston, TX?
2) What types of crime constitute different hot spots?
3) What are the crime trends for the city over time (2018-2020, annual and seasonal)?
4) What factors, variables, or indicators have a spatial relationship with crime hot spots (namely
bars, other land uses)?
5) Does crime increase or decrease before and after natural disaster such as tropical storms and
hurricanes?
5.1.1 Crime hot spots located within the city of Houston, TX?

The highest concentration of crime hotspots occurred in the downtown and midtown
districts of Houston, TX. This trend was found across all crime types. The category 5 and
category 4 crime hotspots were found near the intersection of the downtown, midtown and
61

Montrose districts. Category 3 clusters were also found at I-69 and Kirby and along Washington
Avenue in the bar district and southeast of I-69 and 288, at the intersection of Sampson and
Elgin.
5.1.2 What types of crime constitute different hot spots?

The category 5 and 4 crime hotspots occurred mainly in the downtown and midtown
districts for all 4 crime types. Most of the downtown and midtown districts experienced at least
category 3 hotspots. Category 3 hotspots were also found along Washington Avenue. A large
area of category 2 hotspots for all crimes were found in the southern section of the study area,
south of US 90 and east of 288 near I-610. There were a smaller number of category 2 hotspots
for each crime type in the northeast, southwest, and the eastern section of the study. The upper
Kirby district in the western portion of the study also experienced category 2 hotspots with a few
category 3 hotspots at Kirby and I-69.
5.1.3 Crime trends for 2018-2020

5.1.3.1 Assault Crime Hotspots
In all three years, there was category 5 and 4 clusters near the intersection of the
midtown and downtown districts on I-45. An additional category 5 and a category 4 cluster
popped up in the Montrose District in 2019. The northern section north of I-10 and these areas
east of I-69 saw category 3 and 2 clusters in all 3 years. The southern section south of US 90 and
east of 288 contained category 3 and 2 clusters in all three years. The intersection of I-610 and
US 90 experienced category 3 and 2 clusters in all three years as well.
Four areas saw hotspots in this just one year (2019). For example, category 4, 3 and 2
hotspots occurred at Kirby and Washington in 2019, but not in 2018 or 2020. At Kirby and I-69
in the upper Kirby district, a category 3 hotspot and large category 2 hotspot occurred in 2019
62

and was connected to clusters in the Montrose District. The northwest section along Nineteenth
Street experienced a category 2 cluster in 2019 and at Almeda and I-610, a large category 2
cluster occurred in 2019. Category 2 clusters occurred in two of the three years at Bellfort and
MLK, Canal and US 90, and along Holcombe north of US 90.
5.1.3.2 Burglary Crime Hotspots
All three years experienced category 5, 4 and category 3 burglary hotspots near the
intersection of the downtown and midtown districts. The 2019 category 5 and 4 hotspots spread
into the Montrose District. A large category 2 covered a majority of the downtown, midtown and
Montrose districts in all three years. The upper Kirby district saw category 3 and category 2
hotspots in all three years. Canal and 90 experienced a category 2 hotspot in all three years.
Washington and Kirby experienced category 2 hotspots in all three years. Lockwood and
Collingsworth in the northern section of the study area saw category 2 hotspots in all three years
and the southern section south of US 90 and east of 288 experienced numerous category 2
clusters in all three years. The areas near the I-610 and US 90 intersection experienced category
3 and category 2 hotspots in all three years as well.
Three unique areas in the south-central, southwest and northern section north of Fulton
and Morris experienced small category 2 clusters in 2019. The I-45 at US 90 and the I-45 and I-
610 experienced category 2 clusters in two of the three years. The I-10 and US 90 saw small
category 2 clusters in two of the three years, and Fulton and Morris also experienced small
category 2 clusters in two of the three years.
5.1.3.3 Robbery Crime Hotspots
All three years saw a category 5 and category 4 hotspot next to the I-45 as it crossed the
downtown and midtown districts. In 2019, large category 3 and 4 clusters expanded and covered
63

a majority of the downtown, midtown and Montrose districts. The southeast section on I-45 at
US 90 and the I-610 saw category 2 clusters in all three years, and the northern section
experienced small category 2 clusters north of I-10 and east of I-69 in all three years. The
southern section north of I-610 and east of I-45 saw large category 2 hotspots in all three years.
The I-610 and US 90 intersection and I-610 and Almeda saw large category 2 and a
small category 3 cluster in two of the three years respectively, Fulton and Morris saw a category
2 cluster in two of the three years and south of the I-610 along Cullen, two small category 2
clusters occurred in two of the three years. Washington Ave experienced a large category 2 and a
category 3 cluster in 2019 that did not occur in the other two years. Similarly, the upper Kirby
district saw a category 3 cluster within a large category 2 cluster that extended to the Montrose
district in 2019, but not in the other two years. The upper Kirby district hotspot did not occur in
the other two years. The area near Canal and US 90 saw a category 2 cluster in 2019 as well, but
not in the other two years.
5.1.3.4 Theft Crime Hotspots
Both 2019 and 2020 saw category 5 and 4 hotspots near the intersection of the
downtown, midtown and Montrose districts. In 2019, a large category 4 cluster covered a
majority of the downtown, midtown and Montrose districts. This same region experienced a
category 2 cluster in 2020. In both years, category 3 and 2 clusters occurred at Kirby and I-69.
Two category 2 clusters occurred north of I-10 and just east of I-610 near Westheimer in the
western portion of the study area. Washington Avenue experienced a category 3 cluster and a
large category 2 cluster in both years. At the I-610 and 90, category 2 and 3 clusters occurred in
both years. In the northwest section, category 2 clusters were present along 19
th
in both years.
South of 90 and east of 288 in the southern section, category 2 clusters were found only in 2019.
64

Similarly, category 2 clusters occurred along the I-45 and at Canal and US 90 in the southeastern
section of the study area in 2019, but not in the other years.
These findings suggest that law enforcement should expect to see the highest
concentrations of all crime types near the intersection of the midtown and downtown districts
near I-45. The eastern Montrose district also showed high densities for almost every crime type.
The Washington Ave bar district showed category 3 and 2 hotspots for all crime types. The
upper Kirby district at Kirby and I-69 also displayed category 3 and 2 hotspots for almost every
crime type. The allocation of more police resources and potential policy changes by city officials
could help to reduce the crime levels of these areas in the future.
5.1.4 2020 Land Use Types and Optimized Crime Hotspots
One objective of the project was to better understand the relationship between land use
type and crime hot spots. This was done using Kernel Density Analysis and Optimized Hotspot
Analysis. The land use type attributes were selected from within the optimized crime hotspots
that experienced Gi Bin +3. This represents a 99% confidence level as to whether or not a crime
hotspot is present.
The sum of the area (in acres) of each land use type for each optimized crime hotspot was
calculated. The discovery that a certain land use type accounted for a majority of the total area of
the optimized Gi Bin level +3 does not necessary mean that land use is the cause of the crime
hotspot. The next few paragraphs report the area percentages of each of the land use types
calculated from total land use acres.
The 2020 Assault Optimized Hotspots occurred in the midtown and downtown districts,
which were primarily single-family housing (48%). Another crime hotspot occurred in the
southern portion of the study, which was comprised of single-family housing (68%) and
65

undeveloped land (24%). In the northeastern section of the study area, another crime hotspot was
dominated by single-family housing (61%) and undeveloped land (33%). Thus, the study
suggests that single-family housing made up the largest land area in acres for assault hotspots.
Just because single-family housing units accounted for the largest land use, does not mean that
single-family housing units are the cause of crime hotspots, but rather that this land use is
disproportionately represented in the assault Gi Bin +3 areas.
The 2020 Burglary Optimized Crime Hotspots occurred in the midtown, downtown,
Montrose and part of the upper Kirby district. The land use associated with these crime hotspots
were single-family housing (55%), multi-family housing (17 %), undeveloped land (17%),
commercial (5%) and industrial (3%). Another crime hotspot occurred in the southwestern
portion of the study at US 90 and the I-610. The land use for this crime hotspot are single-family
housing (75%), commercial (11%), and undeveloped land (6%). Again, single-family housing
represented the largest land use in the burglary Gi Bin +3 crime hotspots.
The 2020 Robbery Optimized Crime Hotspots occurred at the intersection of the
downtown and midtown districts. This crime hotspot spilled into the eastern portion of the
Montrose District. The land use types for this crime hotspot were dominated by single-family
housing (47%), multi-family housing (29%), and undeveloped land (20%). Single-family and
multi-family housing represented the largest land use area in acres for robbery Gi Bin +3 crime
hotspots. This does not necessary mean that single-family and multi-family housing are the cause
of robbery hotspots.
The 2020 Theft Optimized Land use occurred in the downtown, midtown, Montrose, and
Kirby districts. This large crime hotspot extended into the bar district along Washington Avenue.
The dominant land use types for this crime hotspot were single-family housing (55%), multi-
66

family housing (17%), commercial (5 %), and industrial (3%). A second crime hotspot occurred
in the western portion of the upper Kirby District dominated by single-family housing (67%) and
multi-family housing (25%). Along I-610 in the western section of the study area, a small theft
hotspot popped up in 2020 in a neighborhood dominated by multi-family housing (54%), and
single-family housing (38%). Overall, single-family housing once again represented the largest
land use type in area (acres) for most of the Gi Bin +3 areas of theft.
5.1.5 Crime and Natural Disasters
Hurricane Harvey occurred on 08/25/17. Crime was analyzed from 07/17 to 10/17 to
include the month before Hurricane Harvey and two months following this major hurricane.
Assault, burglary, and theft all trended downward from 07/17 to 10/17 and robbery trended
upward during this same time period.
Tropical Storm Imelda occurred on 09/17/19. The general trend for the crime rates one
month before and one month after this event was a general decrease in numbers of crimes from
08/19 to 10/19.
Tropical Storm Beta occurred on 09/21/20. The general trend for crime before and after
this event is a general increase in assault, robbery and theft in October compared to August 2020.
Theft increased significantly and burglary was the only crime type to decline from 08/20 to
10/20.
In summary, crime before and after Tropical Storm Beta increased for all types, but for
Burglary, crime decreased for all types before and after Tropical Storm Imelda, and crime before
and after Hurricane Harvey trended downward for all, but robbery. Hurricane Harvey was the
only natural disaster included in this study that occurred before 2018. The two tropical storms
occurred in 2019 and 2020 during the same temporal period of this crime study.
67

5.2 Successes, Challenges, and Failures
As with any large project, one concludes reflecting on aspects of the project they feel
went well and areas that did not go as expected. Understanding these successes, challenges, and
failures can shape future research by the researcher or others.
One success of this study was the geocoding of the study boundary crime addresses. The
ArcGIS World Geocoding Service proved to be time efficient and accurate. For example, there
were 13,944 crime occurrences between the dates of July-August 2020 for the study boundary
edge test. The ArcGIS World Geocoding Service was used to geocode these crime addresses and
97.22% were matched and 0% were left unmatched. The remainder (2.8%) were returned with
two or more of the same addresses or block locations.
Of course, there were also challenges faced in this study. The first was the 100,000+
crime addresses that needed geocoding. This was simply because of the huge number of records
being processed. The second challenge was merging the 1
st
and 2
nd
geocoding attempts with the
CDX WinZip geocoder for Excel into one master file for each year due to the large volume of
data. The third challenge was matching the columns (attributes) of the 3
rd
geocoding attempts in
Geocoding by Smart Monkey with the other geocoding solutions, CDX WinZip and ArcGIS
World Geocoding Service. When matching these attributes, address, city, state and zip needed to
be merged into one attribute column. This was accomplished by creating a concatenated formula.
One way the project failed to meet objectives was the inability at first to geocode the
2018 theft crime occurrences. Given more than 18,000 thefts, there were simply too many crime
sources given the data processing capabilities using the default settings. The solution was to
remove two attributes (theft from building and theft of motor vehicle parts or accessories) from
68

the 2018 Theft records. This reduced the number of 2018 thefts to a more manageable 12,597
occurrences.
The biggest challenge of the data management part of project was merging addresses for
each geocoding attempt. As previously discussed, the (successful) initial attempt was saved in
Excel, the unsuccessful addresses (those that were not located) were rerun using the ambiguous
check box tool, and the final batch was merged using Geocoding by Smart Monkey. While it was
challenging to remove the unsuccessful attempts, they were later added to the spreadsheets once
the latitude and longitude coordinates were located using Geocoding by Smart Monkey. Again,
the main challenge was the amount of data (100,000+ crime addresses). The process would have
been more efficient if the Esri Geocoding Service had been used for all of the 2018-2020 crime
addresses. CDX WinZip geocoding and Geocoding by Smart Monkey were utilized as a test of
effectiveness.
The tools used in this project were adopted optimistically, with the knowledge that their
effectiveness would be better understood after they were put to the test. The CDX WinZip
geocoding tool was used with the 2018 and 2019 data to test the effectiveness of a geocoding
technique other than Esri’s World Geocoding Service. It proved to be effective, but not as
efficient as Esri’s World Geocoding Service. Esri’s World Geocoding Service worked better
given the sheer volume of data being geocoded, as evidenced by the efficient geocoding of the
2020 Edge Effect crimes. CDX WinZip is a powerful tool for geocoding when each batch of
addresses contains a few thousand addresses.
The kernel density tool proved effective, despite the large amount of data. It was still able
to distinguish points from one another, despite their close proximity, and create density hotspot
maps. The Kernel Density Tool was successful using the default settings in ArcGIS Pro. This
69

customized the distance band and raster grid cell size for each dataset of crime points based on
the number of crime points. The default settings accounted for the number of crime points and
the size of the area to create customized settings for each dataset. Having learned a great deal
from this project, future analysis of this data could involve the manipulation of the defaults to
experiment with the implications of manipulating the data with carefully chosen rather than
default inputs.
The exception to the success of the kernel density analysis was the 2018 Theft data,
where there were too many occurrences (18,962) in close proximity, more than any other crime
type or year. While there were attempts to create distance bands and parameter settings for the
2018 theft hotspots at 18,962 crime occurrences, it did not prove successful. The solution was to
remove two attributes from the 2018 theft data and reduce the number of crimes to 12,597.
5.3 Implications for Law Enforcement
The intended audience of this project was law enforcement personnel and authorities in
the Houston area. As previously discussed, knowing past and current crime hot spots can help
law enforcement with planning and preparing for police patrols. This study strongly suggests that
police should focus their efforts more on the downtown and midtown districts, as the primary
category 5 and category 4 hotspots occurred at the intersection of the downtown, midtown and
Montrose districts. These types of areas could be allocated more law enforcement funding and
personnel, such as adding a police station and/or additional police resources.
In addition to law enforcement focusing on the areas with the highest crime, being able to
identify areas of moderate to high crime (category 3 and 4 hotspots) can allow measures to
prevent these areas from getting worse. For example, Kirby Drive and I-69 saw medium sized
category 4 and 3 clusters. Southeast of I-69 and 288, another category 4 and 3 cluster appeared at
70

the intersection of Sampson and Elgin. These areas are critical places in which law enforcement
could and should take preventive measures.
Law enforcement organizations could learn from these findings and prioritize their
resources accordingly. By changing city policy and providing more police resources in critical
areas, the growth of crime in Houston could be curtailed. The category 4 and category 5 hotspots
could be changed into category 3 hotspots. More policing could also prevent category 2 and 3
hotspots from turning into category 4 and 5 hotspots. And were this study replicated in the
future, the impacts of these changes may be temporally and spatially evident.
5.4 Future Research
This study looked at changes in crime and crime types over time in the inner 610 loop of
Houston, TX. It also looked at related variables that may help better understand crime
distributions. However, in the future, a number of changes could be made to increase efficiency
with respect to data management, visualizations, and to explore related variables in more depth.
A new approach and method could involve more police input and participation to make
the data more usable to law enforcement on-the-ground. In particular, the needs and opinions of
law enforcement should be taken into account, and/or their personal experiences could be
incorporated. For one, law enforcement agencies are likely familiar with which types of crimes
are committed in which types of land use zones. This type of knowledge could be mapped using
participatory GIS techniques and be used to ground-truth the data generated through the spatial
analysis methods. Another way in which use the input of law enforcement officers would be with
respect to the resources needed to enforce and address different types of crime. While some types
may require more human-power and be more risky, others are less so. The ability of law
enforcement agencies to address certain types of crime more efficiently than others could lead to
71

further suppression of high and medium crime hotspots for all crime types. These types of
considerations could inform future studies to focus on certain crime types, areas, and time
periods while eliminating the areas of the study that law enforcement believes to be less helpful.
With respect to data management, most suggestions for future research are meant to make
workflows more efficient. Should a researcher seek to conduct a similar study, they must take
three criteria into account: population, time period, and crime rates, also known as counts. The
higher each of these are, the more bandwidth and more time the study will take. Based on the
scope of their project, a future researcher should carefully consider the priorities of the study and
allocate resources accordingly.
To start, Esri’s World Geocoding Service could be used for all of the geocoding tasks. A
master spreadsheet would not have to be created to match all attributes and columns from
different geocoding services such as CDX WinZip and Geocoder by Smart Monkey. This would
reduce the number of geocoding services utilized from three to just one. Another improvement
would be a faster and more efficient method of geocoding the crime addresses. A workflow
could also be created in ArcGIS Pro and/or Model Builder to enhance and accelerate the
geocoding process.
With respect to visualization techniques, crime trends could be better explained using a
Space Time Cube. Space Time Cubes allow the viewer to understand crime trends over periods
of time using 3D visualization. This can help determine new hotspots, consecutive hotspots,
intensifying hotspots, persistent hotspots, sporadic hotspots and historic hotspots. Techniques
from behavioral geography and cognition could be used to understand what types of
visualizations are clearer to a viewer.
72

Future studies could also look at the role of disruptions on crime. Disruptions could
include hurricanes and other natural disasters, or epidemiological emergencies like the COVID
pandemic. These types of disruptions can impact crime rates and types, including migration, lack
of basic resources and general scarcity. Looking at the differences in crime rates either within or
between cities before and after a disruption may enable disaster agencies to be better prepared.
Understanding the demographics of cities that are more or less resilient to crime after disasters
could help general readiness.
More broadly, the techniques used in this study could be applied to other cities that
experience similar impacts due to rising crime rates. The GIS techniques used in this study are
not exclusive to Houston, although each city is likely to have its own unique crime data attributes
and challenges. A study based on the needs of another region could use similar techniques to
achieve different objectives. For example, a study of an area with traffic challenges could
provide optimized routes from police stations to crime hotspot locations. Perhaps predictive
models could be created to anticipate the development of hotspot types and locations in the
future.
This project sought to use spatial data to create up-to-date analysis and visualization of
crime types and frequency in Houston, Texas. The process included the careful geocoding of a
huge volume of crime data, and explains the types of accommodations that were made to achieve
the project’s goal. Land use types and bus stops were considered as possible related or
explanatory variables. The relationships between crime rates before and after natural disasters
were examined. Ultimately, the project provides analysis and visualizations that could be of use
to law enforcement agencies and personnel. These techniques could be replicated or built upon in
different cities to achieve similar or different objectives. This research seeks to improve the
73

efforts of law enforcement in reducing crime, thus improving the quality of life for urban
residents.

74

References
Block, R., and Block, C.R. 1995. “Space, place and crime: hot spot areas and hot places of
liquor-related Crime” in J. E. Eck and D. Weisburd (eds.), Crime and Place. Crime Prevention
Studies, 4. Criminal Justice Press: Monsey, NY. 147-185.

Canter, P. 2006. “Crime Analysis in Baltimore County, Maryland. Baltimore County Police
Department’s Crime Analysis Unit.”

City-Data.com. 2019. “Crime rates in Houston, Texas (TX): Murders, Rapes, Robberies,
Assaults, Burglaries, Thefts, Auto Thefts, Arson, Law Enforcement Employees, Police Officers,
Crime Map”. Accessed 12 October, 2021. http://www.city-data.com/crime/crime-Houston-
Texas.html.

City of Houston GIS. 2021. “Transportation Complete Streets”, COHGIS Data Hub, Accessed
03 October 2021. https://cohgis-mycity.opendata.arcgis.com/search?collection=Dataset.

City of Houston Police Department. 2021a. “2018 Crime Data”. Accessed 15 October, 2021.
http://www.houstontx.gov/police/cs/crime-stats-archives.htm.

City of Houston Police Department. 2021b. “Monthly Crime Data by Street and Police Beat”.
Accessed 05 October, 2020. https://www.houstontx.gov/police/cs/Monthly_Crime_Data_by_
Street_and_Police_Beat.htm.

Cohen, J., Gorr, W.L., and Olligschlaeger, A.M. 1993. “Modeling Street-Level Illicit Drug
Markets”. Working Paper 93–64. Pittsburgh: H. John Heinz III School of Public Policy and
Management, Carnegie Mellon University.

Curry, G.D., and Spergel I.A. 1988. “Gang homicide, delinquency, and community”.
Criminology 26, no. 3: 381–406.

Esri. 2015. “Analyzing Violent Crime”. Esri. Analyzing violent crime, workflow—Analytics |
Documentation (arcgis.com). Accessed April 25, 2021.

Esri. 2021. “Hot Spot Analysis (Spatial Statistics)”. Esri. https://pro.arcgis.com/en/pro-app/tool-
reference/spatial-statistics/hot-spot-analysis.htm. Accessed April 18, 2021.

Gorr, W., and Olligschlaeger, A. 2002. “Crime Hot Spot Forecasting: Modeling and
Comparative Evaluation”. (Draft Final Report.) Washington, DC: Office of Justice Programs,
National Institute of Justice.

Groff, E.R., and La Vigne, N. 2002. “Forecasting the future of predictive crime mapping.” In N.
Tilley (ed.), Analysis for Crime Prevention 13: 29-58. Criminal Justice Press: Monsey, NY.

75

Levine, N. 2015. “CrimeStat: A Spatial Statistics Program for the Analysis of Crime Incident
Locations” (v 4.02). Ned Levine & Associates, Houston, Texas, and the National Institute of
Justice, Washington, D.C. August. 2015.

Levine, N., Wachs, M., and Shirazi, E. 1986. “Crime at bus stops: A study of environmental
factors”. Journal of Architectural and Planning Research 3, no. 4: 339-361.

Li, Z., Liu, Q., Tang, J., and Deng, M. 2018. “An adaptive method for clustering spatio‐temporal
events. Transactions in GIS 22, no. 1: 323-47. https://doi.org/10.1111/tgis.12312.

Nakaya T., and Yano, K. 2010. “Visualising Crime Clusters in a Space‐time Cube: An
Exploratory Data‐Analysis Approach Using Space‐time Kernel Density Estimation and Scan
Statistics.” Transactions in GIS 14, no. 3: 223-39. https://doi.org/10.1111/j.1467-
9671.2010.01194.x.

Paisley, L. 2017. “USC Spatial Sciences Students Analyze Crime Patterns in L.A.” University of
Southern California News. June 28, 2017. Accessed on November 15, 2021.
https://news.usc.edu/115324/usc-spatial-sciences-students-analyze-crime-patterns-in-l-a/.

Roncek, D.W., and Maier, P. 1991. “Bars, Blocks, and Crimes Revisited: Linking the Theory of
Routine Activities to the Empiricism of ‘Hot spots’”. Criminology 29, no. 4: 725–53.
http://dx.doi.org/10.1111/j.1745-9125.1991.tb01086.x.

Sherman, L.W., Gartin, P., and Buerger, M.E. 1989. “Hot Spots of Predatory Crime: Routine
Activities and the Criminology of Place. Criminology 27, no.1: 27–56.

Shiode, S. 2011. “Street‐level Spatial Scan Statistic and STAC for Analyzing Street Crime
Concentrations.” Transactions in GIS 15, no. 3: 365-83. https://doi.org/10.1111/j.1467-
9671.2011.01255.x.

Spelman, W. 1995. "Criminal careers of public places." In: D. Weisburd and J. Eck (eds.), Crime
and Place. (Crime Prevention Studies, no. 4: 115-144). Monsey, NY: Criminal Justice Press.

United States Census Bureau. “Houston, TX facts”. Accessed from
https://www.census.gov/quickfacts/houstoncitytexas on November 12, 2021.

Wheeler, A.P., Steenbeek, W., Andresen M.A. 2018. “Testing for similarity in area‐based spatial
patterns: Alternative methods to Andresen's spatial point pattern test.” Transactions in GIS,
22: 760–74. https://doi.org/10.1111/tgis.12341.

A p p e n d i x A . B u s S t o p D e n s i t y a n d 2 0 2 0 T h e f t H o t s p o t s
7 6
2 0 2 0 T h e f t
C r im e H o t S p o t s
L o w
M e d iu m
H ig h
V e r y H ig h
B u s S t o p D e n s it y
V A L U E
L o w
M e d iu m
H ig h
V e r y H ig h
1 i n c h e q u a l s 3 . 5 m i l e s 1 i n c h e q u a l s 3 . 5 m i l e s
H a r r i s C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
1 0
4 5
6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
1 0
4 5 6 9
6 9
9 0
6 1 0
4 5
9 0
9 0
6 9
6 1 0
( a ) B u s S t o p D e n s i t y ( b ) 2 0 2 0 T h e f t H o t S p o t s
2 8 8
2 8 8
6 1 0
6 1 0
6 1 0
±
T o p 5 0 B a r s
H o u s t o n S t r e e t s
0 3 0 6 0 1 5 M ile s
±
H a r r i s C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
1 i n c h e q u a l s 1 m i l e
7 7 A p p e n d i x B . T h e b a r l o c a t i o n s w i t h i n t h e S t u d y D i s t r i c t s .
1 0
4 5
6 9
4 5
1 0
6 9
2 8 8
6 1 0
M o n t r o s e D i s t r i c t
U p p e r K i r b y D i s t r i c t
M i d t o w n
D T
W a s h i n g t o n A v e
W e s c o t t
K i r b y D r i v e
K i r b y D r i v e
A p p e n d i x C . M e t r o B u s S t o p D e n s i t y
1 i n c h e q u a l s 3 . 5 m i l e s
1 0
4 5 6 9
4 5
9 0
6 1 0
7 8
1 i n c h e q u a l s 3 . 5 m i l e s 1 i n c h e q u a l s 3 . 5 m i l e s
H a r r i s C o u n t y , T X a n d S t u d y A r e a ( O r a n g e )
1 0
4 5 6 9
4 5
9 0
6 1 0
6 1 0
6 1 0
9 0
6 9
6 1 0
( a ) B u s S t o p D e n s i t y ( b ) M e t r o B u s S t o p s
2 8 8
B u s S t o p D e n s it y
V A L U E
L o w
M e d iu m
H ig h
V e r y H ig h
C O H M E T R O B U S S T O P S
H o u s t o n S t r e e t s
I n t e r s t a t e 6 1 0
±
6 1 0
6 1 0
6 1 0
6 1 0

Abstract (if available)

Abstract In 2018, Houston’s crime rate was higher than the rates in 95% of U.S. cities. Houston’s population is the fourth highest in the nation with more than 2 million people, all of whom are affected by this high crime rate. A better understanding of the spatial and temporal aspects of crime would be useful for law enforcement in protecting the general population. This study analyzed assaults, burglaries, robberies, and thefts in the inner Interstate 610 area of Houston, which is considered downtown. The Houston police department provided crime address data for each crime type from 2018 to 2020. The crime data was geocoded in ArcGIS Pro into point shapefiles and aggregated using counts. The Esri Optimized Hot Spot Analysis and Kernel Density Tools were used to determine crime hot spots for each crime type. The study also explored whether land use type was related to hotspots of certain crimes in the Downtown and Midtown districts of Houston. The study found that the crime hotspots for each crime type occurred mainly in the Downtown, Midtown, and Montrose districts of Houston. Thefts and assaults were higher near the downtown bar district. Theft was also higher near bus stations. The study results could be valuable in helping the police predict and respond to crime hot spots in the future in the Houston area, and the methods used may help the police manage crime in other geographic areas and over different time periods as well.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The spatial effect of AB 109 (Public Safety Realignment) on crime rates in San Diego County

PDF

Testing social disorganization theory on violent crime: a case study on Pueblo, Colorado

PDF

Scenario-based site suitability analysis and framework for biodiversity conservation: agricultural zone, Galapagos Archipelago, Ecuador

PDF

Spatiotemporal analysis of the SLOSH and ADCIRC storm surge models: a case study of hurricane Ida

PDF

A spatiotemporal analysis of environmental risk factors of Lyme disease in the Northeastern United States

PDF

The impact of definition criteria on mapped wildland-urban interface: a case study for ten counties along the Oregon-California border

PDF

Mapping punk music and its relative subgenres

PDF

A spatiotemporal analysis of racial disparity in the distribution of superfund sites within Santa Clara County, California

PDF

Geographic information systems and marketing: a transdisciplinary approach to curriculum development

PDF

Historical temperature trends in Los Angeles County, California

PDF

Light rail expansion in Houston using viable path corridors and least cost path: alternatives for the failed University and Uptown lines

PDF

That sinking feeling: predicting land subsidence in California’s San Joaquin Valley with a spatial regression model

PDF

Creating a flood vulnerability index for Houston, Texas

PDF

Creating Hot Streets: developing an automated approach using ModelBuilder

PDF

Assessing homeless accessibility to community resources in the city of Los Angeles

PDF

A comparative study of ground and satellite evapotranspiration models for southern California

PDF

Estimating at-risk population for lead service lines induced lead exposure and their correlation to socioeconomically disadvantaged neighborhoods in Milwaukee, Wisconsin

PDF

Operational optimization model for Hungry Marketplace using geographic information systems

PDF

Bringing GIS to a small community water system

PDF

Evaluating the MAUP scale effects on property crime in San Francisco, California

Asset Metadata

Creator Shreve, Geoffrey Jacob (author)

Core Title Spatiotemporal hotspots of 2018-2020 crime in Houston, Texas

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Degree Conferral Date 2022-12

Publication Date 09/17/2022

Defense Date 08/30/2022

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

Tag 2018,2019,2020,assault,Burglary,Crime,Houston,OAI-PMH Harvest,Robbery,space,Texas,Theft,Time

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Ruddell, Darren (committee chair), Duan, Leilei (committee member), Wilson, John (committee member)

Creator Email geoff_shreve@sbcglobal.net,shreve@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111995946

Unique identifier UC111995946

Legacy Identifier etd-ShreveGeof-11215

Document Type Thesis

Format application/pdf (imt)

Rights Shreve, Geoffrey Jacob

Type texts

Source 20220917-usctheses-batch-981 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

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

Repository Name University of Southern California Digital Library

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

Repository Email cisadmin@lib.usc.edu