The Impact of COVID-19 on Crime: A Study from the

Spatial-temporal Perspective in the Montgomery County, AL

Long Ma

and Connor Y. H. Wu

Department of Computer Science, Troy University, Troy, AL, U.S.A.

Department of Geospatial Informatics, Troy University, Troy, AL, U.S.A.

Keywords: COVID-19, Crime Trend, Spatial-temporal, Machine Learning, Montgomery.

Abstract: The policies curbing the spread of COVID-19 can influence the chance of committing a crime. This study

aimed to investigate the impacts of COVID-19 on the spatial and temporal patterns of crime in Montgomery

City, AL, by wavelet analysis, spatial point test, and machine learning tools. We obtained the crime case

records between January 1, 2015 to March 12, 2021 from the police department in the City of Montgomery,

and we downloaded demographical data from the U.S. Census. Results show that the overall crime rate in

Montgomery decreased during the COVID-19 pandemic. However, crime rates would increase in a shorter

time than COVID-19 confirmed cases when the social activities increased. Meanwhile, spatial distributions

of simple assault, burglary, and vehicle theft had clustered in Montgomery business and shopping areas. These

findings are helpful for the police institution in preventing and minimizing crimes as new COVID-19 variants

emerge in the future.

1 INTRODUCTION

The COVID-19 pandemic has spread globally and

impacts every aspect of people’s daily life (Boman &

Mowen, 2021). Governments have implemented

stay-at-home orders and social distancing

requirements to curb the spread of the COVID-19

virus within communities (Piquero, et al., 2021).

People have been requested to limit social contacts,

avoid social gatherings, close schools, and stop

unnecessary business activities (Koh, 2020). Crimes,

therefore, have been changed. For instance, overall

crimes are reported to drop sharply, with approximate

37% worldwide (Boman & Mowen, 2021), 35% in

the United States (Abrams, 2021), 41% in the United

Kingdom (Halford, et al., 2020) during the COVID-

19 pandemic. However, the impacts are varied on

different types of crimes. For instance, some crimes

decreased (e.g., burglary and robbery), some crimes

increased (e.g., domestic violence), some crimes had

no changes (e.g., assault-battery) in Los Angeles and

Indianapolis (Mohler et al., 2020).

This study aims to investigate the impacts that

COVID-19 has had on the spatial and temporal

patterns of crime in Montgomery City, AL, through

spatial and temporal crime analysis approaches.

Specifically, we analyzed the temporal pattern

between COVID-19 and crimes through wavelet

analysis. Then, we explored the spatial pattern

changes of different crimes on March 13, 2020 (the

date with the first COVID-19 confirmed case

reported) through March 12, 2021 (one year after the

first confirmed case) by using the spatial point pattern

test (SPPT) and utilized the local Moran’s I to identify

regional clusters and local spatial outliers.

2 RELATED WORKS

Some studies have analyzed the impact of the

COVID-19 pandemic on the spatial and temporal

distributions of crimes. For instance, Yang, Chen,

Zhou, Liang, and Bai (2021) found that the

distributions of crime in Chicago significantly

changed in 2020, with local changes in theft, battery,

burglary, and fraud displaying an aggregative cluster

in the downtown area. However, because of the

geographical variation, whether their results in a big

city like Chicago can be applied to a middle-sided city

like Montgomery is unknown. Additionally, existing

studies have mainly considered the crime and

COVID-19 data a few months after the COVID-19

occurred in February and March 2020 ( Mohler et al.,

2020). The holiday months at the end of 2020 with

532

Ma, L. and Wu, C.

The Impact of COVID-19 on Crime: A Study from the Spatial-temporal Perspective in the Montgomery County, AL.

DOI: 10.5220/0010856700003116

In Proceedings of the 14th International Conference on Agents and Artiﬁcial Intelligence (ICAART 2022) - Volume 3, pages 532-539

ISBN: 978-989-758-547-0; ISSN: 2184-433X

the highest numbers of confirmed cases occurred in

many states, and the months when many people got

vaccinated are not considered.

3 MATERIAL AND METHODS

3.1 Study Area and Data

Montgomery is the second-largest city in Alabama,

with a population of 205,764 according to the 2010

census. Montgomery was divided into 199 block

groups, the smallest geographic area for which the

U.S. Census collects and tabulates decennial census

data. In Montgomery, the first case of COVID-19 was

reported on March 13, 2020. The number of

confirmed cases increased and reached 22,232, of

whom 526 had died in its county as of March 12,

2021. During the COVID-19 period, the Stay-at-

Home order was issued on April 3, 2020, which was

canceled on April 30, 2020. The Safer-at-Home

order, which required wearing a mask and social

distancing, was issued on April 30, 2020 and

withdrawn on April 4, 2021.

We obtained the crime case records with date,

time, location (X and Y), crime description, and type

between January 1, 2015 to March 12, 2021 in

Montgomery City from the police department from

the City of Montgomery. Demographical data

(including income and race) at the block group level

used in the analysis were downloaded from U.S.

Census. A block group is a subdivision of a census

tract and consists of blocks. One block group usually

has between 250 and 550 housing units (The United

States. Bureau of the Census, 1994). We used block

groups in the spatial analysis because they are the

smaller geographic areas than census tracts and Zip

codes, and they have a higher number of housing

units for the sample size than blocks.

3.2 Proposed Methods

3.2.1 Wavelet Analysis

The wavelet approach is a proper statistical method

that has been applied in various academic fields (Wu

& Loo, 2017). Because crime and climate data are

constantly varying time series variables influenced by

factors, such as changes in the physical environment,

related laws and policies, and criminal demographics,

we applied wavelet coherency analysis to examine the

possible non-linear and non-stationary connection

between environmental factors and crime rates.

Wavelet coherence allows us to explore the correlation

between two non-stationary signals at a given time and

frequency. Also, we conduct phase analyses to figure

out how the signals are associated. The phase

difference [i.e., in-phase (positively correlated) or out

of phase (negatively correlated)] indicates their

association. Before the wavelet analysis, we normalize

the time-series data by using the formula:

𝑋



𝑋−𝜇

𝜎

which X is the time-series data, µ is the mean of X,

and σ is the standard deviation of X.

3.2.2 Spatial Point Pattern Test

This study used SPPT to exam changes or differences

in two different spatial patterns of points based on the

unit area (Andresen, 2009), and SPPT GUI is an open

source software in GitHub (https://github.com/

nickmalleson/spatialtest, accessed on March 6, 2021).

This study applied SPPT to compare the similarity of

spatial distribution patterns of crime in 2020, 2019,

2018, 2017, and 2016 and investigated the local

changes in crime on the level of block groups to

explore whether the pandemic had affected the spatial

distribution of crime. The global S-index (the index

of similarity used to confirm the similar degree of two

spatial point patterns) value of census tracts is larger

than community areas. The global S-index value of

blocks is the largest, yet it needs too much computing

time. Therefore, we took the Montgomery block

groups as the unit area of SPPT.

SPPT includes three parameters: the number of

iterations, sample size, and confidence interval. The

number of iterations is the number of repeated

samplings of the test dataset. The sample size is the

percent size of the test dataset randomly sampling, and

confidence interval based on the test dataset is used to

determine the similarity significance of the two

samples. According to existing spatial analysis

(Andresen, et al., 2017), the number of iterations was

set to 200, the sample size to 85 percent, and the

confidence interval to 95 percent in the analysis. SPPT

identifies the spatial point patterns that diverge in areas

and aggregates the similarities at the local level into a

global index (Wheeler, et al., 2018). Taking the

calculation of the global S-index of crimes in 2020

compared with 2019 as an example, the test can be

described as follows: 1) Adopt crimes in 2020 as the

base dataset and crimes in 2019 as the test dataset (the

test detects spatial pattern variations of the base dataset

relative to the test dataset); 2) Randomly sample 85%

of the test dataset 200 times, and then calculate the

percentage of crimes in census tracts to generate a 95%

The Impact of COVID-19 on Crime: A Study from the Spatial-temporal Perspective in the Montgomery County, AL

533

confidence interval, and 3) Determine whether the

percentage of the primary data in the census tracts falls

into the confidence interval, obtain the value of the

local S-index, and calculate the global S-index.

There are two critical values. One of them is the

global S-index. Another is the local index, which is

applied to identify statistically significant changes on

the micro-scale (local changes). The local S-index has

three values (−1, 0, 1), which means the base dataset

is lower than, similar, or higher than the test dataset

in a spatial unit, respectively. The global S-index

value is the count of the local S-index, which equals

zero, and then divides the number of all spatial units.

The value of the global S-index ranges from 0 (no

similarity) to 1 (perfect similarity), and 0.80 is used

as the threshold to indicate that two spatial point

patterns are similar. Furthermore, we used Moran’s I

to explore the spatial autocorrelation of local changes

to observe the epidemical impact on local areas.

3.2.3 Crime Trend Discovery

As the continuing pandemic of the COVID-19, many

of us face a long-term impact on our lives. Since the

COVID-19 pandemic and the resulting economic

recession have negatively affected many people’s

financial situation and mental health, people’s lives

and relations have dramatically changed over the past

years. Our research expects to discover the local

crime trends in Montgomery during the COVID-19.

As a branch of artificial intelligence, machine

learning is used in many areas that enable a computer

to learn without being explicitly programmed.

Machine learning has already been applied to

COVID-19 related research in these years; however,

few research studies are focused on criminal justice.

Thaipisutikul, et al., in 2021 proposed a framework

to classify the illegal and violent activities from

online Thai news during the COVID-19 pandemic. In

our research, the hidden correlations between crime

trends and the effect of the pandemic are challenging

to disclose; therefore, we apply machine learning to

reveal the relationships between the number of

reporting crimes and the number of COVID-19 cases

in Montgomery from March 13, 2020 to March 13,

2021. We rely on two types of COVID-19 cases in

Montgomery; one is the number of confirmed cases,

and the other is the number of deaths caused by

COVID-19.

Our research adopts one of the most widely used

machine learning methods, supervised learning.

Under supervised learning, the data set is well

labeled; every data instance is tagged with a pre-

defined category. Therefore, each data example is a

pair that consists of an input (X) and corresponding

output (Y). During the training phase, labeled data

has been fed into a machine learning algorithm, which

produces a mapping function to map the input data

with the output value. The generated mapping

function can map the unseen examples to a label in

the test step. Some widely used metrics can be applied

to evaluate the performance of the inferred function

used in the test data. In a word, supervised learning

aims to learn a mapping function from the labeled

data and discover the relationship between the input

and data output. Supervised learning can further be

categorized into classification and regression.

Because our data output is continuous, we will apply

the regression to our research.

To investigate the relationship between the crime

trends and the COVID-19 pandemic, we examine

how the confirmed cases or/and deaths affect the

number of total crimes and the number of individual

crime types in Montgomery. In our research study,

the number of crimes in Montgomery is used as the

output (Y); the number of confirmed cases or/and

deaths is used as input of data (X). We use two state-

of-art machine learning algorithms to fulfill our tasks,

the Support Vector Regression (SVR) and the Linear

Regression. SVR uses the same strategy as Support

Vector Machine (SVM) but is used for regression

tasks. Linear regression models a linear relationship

between input variables and the single output.

Additionally, we use the Root Mean Square Error

(RMSE) method to evaluate the regression

performance of machine learning models.

4 RESULTS AND DISCUSSION

There were 22,944 criminal cases in Montgomery

County from March 13, 2020 to March 13, 2021.

Compared to the pre-pandemic year of 2019 (i.e.,

March 13, 2019 to March 12, 2020), total crime fell

by around 8.26%. For different types of crime, the

numbers of burglary, larceny, traffic violations,

vehicle theft, simple assault, and suicide decreased by

27.47%, 25.30%, 19.72%, 13.43%, 2.91%, and

1.52%, while the numbers of manslaughter,

aggravated assault, robbery, murder, domestic crime,

and rape increased by 42.24%, 38.36%, 18.86%,

14.43%, 6.02%, and 4.41% (Table 1). The number of

crimes of different crime types and the rate of crime

change one year after (March 13, 2020 to March 12,

2021) and before (March 13, 2019 to March 12, 2020)

the first COVID-19 confirmed case are illustrated in

Table 1. The symbol "+" means an increment, and

the symbol "−" refers to a decrement.

ICAART 2022 - 14th International Conference on Agents and Artiﬁcial Intelligence

534

Table 1: The number of crimes of different crime types and the rate of crime change.

Crime Type

Number of Crimes in one

year before (March 13, 2019

to March 12, 2020

)

Number of Crimes in one

year after (March 13,

2020 to March 12, 2021

)

Rate of Change

Murde

970 1,110 +14.43%

Manslau

hte

490 697 +42.24%

Rape 68 71 +4.41%

Aggravated Assault 73 101 +38.36%

Simple Assault 3784 3674 -2.91%

Robber

350 416 +18.86%

Bur

lar

5101 3700 -27.47%

Larcen

4249 3174 -25.30%

Vehicle Theft 1117 967 -13.43%

Domestic Crime 5120 5428 +6.02%

Suicide 261 257 -1.53%

Traffic Violations 71 57 +19.72%

Table 2: The means and standard deviations of crime rates (per 1,000 people) in the communities with different population

density and races in one year after (March 13, 2020 to March 12, 2021) and before (March 13, 2019 to March 12, 2020) the

first COVID-19 confirmed case.

Year Popu_Density Mean (SD) P-value Race Mean (Sd) P-value

One year

after

Low 0.199 (0.073)

<0.001

White 0.166(0.055)

<0.001

High 0.326 (0.079)

Africa

America

0.351(0.070)

Five

years

before

Low 0.237 (0.055)

<0.001

White 0.185(0.058)

<0.001

High 0.373(0.087)

Africa

America

0.412(0.082)

In Table 2, when we grouped the crime cases by

the population density of the block groups they

located in, the results show that block groups with a

low population density would have a significantly

lower crime rate than the ones with a high population

density [i.e., 0.199 (SD=0.073) vs. 0.326 (0.079).

p<0.001]. The crime rate in the white community was

significantly lower than the rate in the Africa

American communities [i.e., 0.166 (SD=0.055) vs.

0.351 (0.070). p<0.001], even though similar findings

can be found in the past five years.

Table 3 shows the results of the global Moran’s I,

which is generally used to indicate the global spatial

autocorrelation. Values of the global Moran’s I range

from -1 to +1. Values above zero indicate positive

spatial autocorrelation, and values below zero

indicate negative spatial autocorrelation. Moreover,

the significance of the global Moran’s I values can be

transformed to the p-value and Z-score. The p-value

is the significance level of Moran’s I, and the Z-value

is the Moran’s I statistic standard deviation. Table 3

also displays that the p-values of rape, simple assault,

and burglary were less than 0.05. The Z-scores are

greater than 1.96, indicating that crimes' spatial

distributions were significantly autocorrelated. The

global Moran’s I value of simple assault and burglary

Table 3: Global Moran’s I of local changes between 2020

and 2019 in different crimes.

Crime Types

Global

Moral's I

p-Value Z-Score

Murde

-0.055 0.222 -1.222

Manslaughte

-0.086 0.052 -1.941

Rape -0.104 0.018 -2.358

Aggravated

Assault

0.015 0.635 0.475

Simple Assault 0.094 0.020 2.325

Robbery -0.055 0.243 -1.167

Bur

lar

0.101 0.012 2.499

Larcen

-0.006 0.974 -0.032

Vehicle Theft -0.015 0.810 -0.241

Domestic

Crime

0.063 0.113 1.584

Suicide 0.064 0.106 1.617

Traffic

Violations

0.033 0.355 0.924

were larger than 0, proving that local changes in these

two crime types display a positive spatial

autocorrelation. In contrast, the global Moran’s I

value of rape was smaller than 0, suggesting that local

changes in these two crime types exhibit a negative

spatial autocorrelation. The significant

autocorrelations of rape, simple assault, and burglary

The Impact of COVID-19 on Crime: A Study from the Spatial-temporal Perspective in the Montgomery County, AL

535

Figure 1: Continuous wavelet power spectra (left) and global wavelet spectrum (right) of the time series.

verified that the spatial distributions of some types of

crimes are associated with the COVID-19 pandemic.

Turning to wavelet analysis, we first present the

wavelet power spectra of the time series in Figure 1

after minimizing the red-noise bias by dividing the

wavelet power by the period. In our crime rates time

series, there is a 3 to 5-day periodicity band in the

panels of COVID-19 confirmed cases, particularly

during the holiday season (November 2020 to January

2021). A similar band in 3 to 5 days also can be found

from August to November 2020 in the panel of

COVID-19 death cases. In Figure 1, Blue represents

lower power values, while yellow represents high

values. The red curve shows the influence cone

delimiting the region from the edge effect. The

dashed black line indicates the 95% confidence

interval based on 10,000 Markov bootstrapped series.

We compute the phases of criminal cases and

COVID-19 cases (e.g., confirmed and death cases)

and their phase difference to obtain additional

information about the linkage between COVID-19

and crime. Figure 2 reveals that the number of

COVID-19 confirmed cases was out of phase with the

number of crime cases, with a delay of ½ of a quasi-

cycle in the holiday season in 2020. The number of

crime cases leads the number of COVID-19 death

cases by ¼ quasi-cycle between July and October

2020. Blue represents lower power values, while

yellow represents high values. The red curve shows

the influence cone delimiting the region from the

edge effect. The dashed black line indicates the 95%

confidence interval based on 10,000 Markov

bootstrapped series. The number of crime cases, with

a delay of ½ of a quasi-cycle in the holiday season in

2020. The number of crime cases leads the number of

COVID-19 death cases by ¼ quasi-cycle between

July and October 2020.

Figure 2: Phases of crime and COVID-19 cases were

computed in the 3- to 5-day periodic band (cf. Figure 1).

After the SPPT test, all crime types (except traffic

violations) had a global S-index value of less than 0.8.

These values are low between 2019 (base dataset) and

2018 (test dataset), similar to the values between 2020

(base dataset) and 2019 (test dataset). The global S-

index values described that the spatial distribution

trend of crimes is not stable and usually changes

significantly every year. We cannot determine

whether the COVID-19 pandemic has impacted the

spatial distributions of crimes with the global S-index

values. However, results of the local S-index showed

that changes in some regional areas are relatively

ICAART 2022 - 14th International Conference on Agents and Artiﬁcial Intelligence

536

stable. Andresen et al. observed the spatial

characteristics of crimes based on local changes and

proved the importance of smaller spatial units of

analysis before our research. Thus, we investigated

the variation of crimes in local spatial units in the

following part. We subdivided the percentage

difference of spatial units between 2020 and 2019

into several classes when local S-index values are not

equal to zero. Then, we used gradation color symbols

to display the percentage differences in Figure 3.

Figure 3: Differences in the percentages between 2020 and

2019 of different crimes’ spatial distributions based on the

spatial unit when SPPT results are significant.

There is an aggregation region of simple assault,

suicide, domestic crime in the eastern business

district. The aggregation regions of simple assault and

domestic display much growth in crimes in 2020

compared with 2019, and suicide shows a decline in

crimes in 2020 compared with 2019. These results

indicated a significant difference in the spatial pattern

of crimes during the pandemic, and the differences in

space are mainly reflected on the microscopic scale.

The local Moran’s I result displays different

clusters (including high-high clusters, low-low

clusters, low-high spatial outliers, and high-low

spatial outliers). In Figure 4, there are high-high

clusters of simple assault, burglary, and vehicle theft

in east-central Montgomery. We also found that

suicide contains high-high clusters in western

Montgomery. The high-low spatial outliers

representing the higher level of this region than

surrounding areas should be noted.

Figure 4: The local Moran’s I for theft, battery, burglary,

and fraud.

Table 4: RMSE scores for evaluating the performance of

predicting the number of all individual crime types by

utilizing the confirmed cases or/and deaths in Montgomery.

Combined

Cases

Confirmed

Cases Onl

Deaths Only

SVR LR SVR LR SVR LR

Murder 3.73 3.59 3.78 3.6 3.73 3.6

Manslau

hter 1.29 1.29 1.29 1.28 1.29 1.3

e 0.44 0.44 0.44 0.44 0.44 0.44

Aggravated

Assault

0.68 0.66 0.68 0.66 0.68 0.65

Sim

le Assault 3.42 3.43 3.41 3.44 3.44 3.43

Robber

1.63 1.57 1.63 1.57 1.63 1.57

Bur

lar

4.96 5.02 4.95 5.02 4.91 5.03

Larcen

3.61 3.61 3.61 3.62 3.61 3.61

Vehicle theft 1.77 1.74 1.71 1.74 1.88 1.77

Domestic

Crime

5.05 5.1 5.07 5.15 5 5.08

Suicided-

related Crime

0.98 0.97 0.97 0.97 0.98 0.98

Traffic

Violations

1.19 1.11 1.19 1.12 1.19 1.11

Others 4.81 4.75 4.81 4.75 4.78 4.62

The Impact of COVID-19 on Crime: A Study from the Spatial-temporal Perspective in the Montgomery County, AL

537

Table 4 illustrates the RMSE scores calculated for

each crime type, such as murder, robbery, rape, and

so forth. We group and count the same crime type

every day in Montgomery from

March 13, 2020 to

March 12, 2021

. Next, we build machine learning

models to predict the number of individual crime

types via using combined cases or/and deaths on a

specific date. The machine learning algorithms SVR

with kernel “Linear” and Linear Regression are also

adopted in this experiment.

5 DISCUSSIONS

The ongoing COVID-19 pandemic has made a

significant impact on people's activities and daily

lives. This study investigated the changes of 12 types

of crime in Montgomery, AL, over one year after the

first COVID-19 confirmed case reported (from

March 13, 2020 to March 12, 2021) based on spatial

and temporal crime analyses.

Compared with one year before the first COVID-

19 confirmed case, the general crime fell by around

8.26% numbers of some specific crime types

increased, such as manslaughter (42.24%), aggravated

assault (38.36%), robbery (18.86%), murder (14.43%),

domestic crime (6.02%), and rape (4.41%). These

results are similar but different from the findings in

existing research. For instance, Yang et al. (2021)

found that total crime fell by 23.7% in Chicago from

February to June in 2020 (with -34.21% in theft, -

29.11% decrease in fraud, -18.97% in the assault, -

18.15% in battery, -7.53% in robbery, -6.54% in

criminal damage, and 3.13% in the burglary). Nivette

et al. (2021) found that overall crime declined by 37%

following stay-at-home restrictions due to COVID-19

in 27 cities across 23 countries in the Americas,

Europe, the Middle East, and Asia.

One of the contributions this study makes is

investigating the impact of COVID-19 on crime from

the spatial perspective. For instance, we grouped

crime cases based on the population density of the

block group. We found that the crime rate in high

populated block groups was significantly higher than

in low populated block groups. However, their

difference became smaller than their difference in the

past years. Meanwhile, when the criminal cases were

grouped by race, we found that the crime rate in the

block groups dominated by Africa America was

significantly higher than that in the white-dominated

block groups. The difference in crime rates between

African America and white-dominated block groups

also narrowed down compared with the rate

difference in the past five years. These results suggest

that the crime rates decreased in different

communities in the COVID-19 pandemic.

Our wavelet analysis results show a 3 to 5-day

periodicity band in the panels of COVID-19

confirmed cases during the holiday season

(November 2020 to January 2021). Moreover, the

number of crime cases leads to the number of

COVID-19 confirmed cases by 1/2 quasi-cycle. This

result suggests that crime rates would increase in a

shorter time than COVID19 confirmed cases when

the social activities increased (such as stay at home

order ended, reopened economy, back-to-school,

presidential election, holidays). It might be possible

that when the social activities decreased, the virus-

prone environment would be clean. When social

activities resumed, the crime-related actions would

occur immediately, but the virus would take time to

spread in a cleaned environment.

There is an aggregation region of simple assault,

burglary, and vehicle theft in east-central

Montgomery. This location coincides with the

business and shopping mall areas, including famous

shopping malls (e.g., The Shoppers at Eastchase,

Dillard’s, Costco, and Target) and restaurants. It is the

most crowded area in Montgomery and is usually a

hot spot for crimes. As the wave of the COVID-19

pandemic poured in, senior centers, libraries, parks,

and other city services were forced to close.

Moreover, this region contains many commercial

buildings with many employees who began to work

at home during this period. The stay-at-home orders

urged most people to quarantine at home, resulting in

an insufficient flow of visitors here. Previous works

have shown that the contextual characteristics of

different areas would impact some crimes, such as the

fact that the busy streets of the city center can attract

more cases of robbery and theft, and the commercial

land can attract more violence. Unlike these crime

types, the aggregation region of burglaries shows

significant growth, resulting from many closed

business districts and the lack of regulators. The

aggregation region of burglaries suggested that the

police department prevent these crimes from

increasing in this region again in the future.

Through looking into the machine learning

performance in Table 5, the crime trend analysis

reveals the confirmed cases or/and deaths in the

COVID-19 pandemic have noticeable relations with

a few certain crime types, e.g., rape and aggravated

assault. Existing research discovers a rise in sexual

violence during the COVID-19 pandemic in

Bangladesh (Sifat, 2020). Another research study

conducted by Roesch et al., devised violence against

women during pandemic restriction.

ICAART 2022 - 14th International Conference on Agents and Artiﬁcial Intelligence

538

6 CONCLUSIONS

In sum, we obtained the following conclusions: 1)

The overall crime rate decreased in Montgomery

during the COVID-19 pandemic, but some crimes

were very sensitive to some policies or events during

the pandemic, like the number of manslaughters,

aggravated assault, robbery, and murder; 2) crime

rates would increase in a shorter time compared with

COVID19 confirmed cases when the social activities

increased; 3) spatial distributions of simple assault,

burglary, and vehicle theft had clustered in

Montgomery business and shopping areas. These

conclusions are significant for preventing and

controlling crimes when the second wave of the

COVID-19 outbreak in Montgomery, such as which

types of crimes should be focused on, and which

regions should be concerned with crime.

REFERENCES

Abrams, D. S. (2021). COVID and crime: An early

empirical look. Journal of Public Economics, 194,

104344.

Andresen, M. A. (2009). Testing for similarity in area-

based spatial patterns: A nonparametric Monte Carlo

approach. Applied Geography, 29(3), 333-345.

Andresen, M. A., Linning, S. J., & Malleson, N. (2017).

Crime at places and spatial concentrations: Exploring

the spatial stability of property crime in Vancouver BC,

2003–2013. Journal of quantitative criminology, 33(2),

255-275.

Boman, J. H., & Mowen, T. J. (2021). Global crime trends

during COVID-19. Nature Human Behaviour, 5(7),

821-822.

Halford, E., Dixon, A., Farrell, G., Malleson, N., & Tilley,

N. (2020). Crime and coronavirus: social distancing,

lockdown, and the mobility elasticity of crime. Crime

Science, 9(1), 1-12.

Koh, D. (2020). COVID-19 lockdowns throughout the

world. Occupational Medicine, 70(5), 322-322.

Mohler, G., Bertozzi, A. L., Carter, J., Short, M. B., Sledge,

D., Tita, G. E., Brantingham, P. J. (2020). Impact of

social distancing during COVID-19 pandemic on crime

in Los Angeles and Indianapolis. Journal of Criminal

Justice, 68, 101692.

Piquero, A. R., Jennings, W. G., Jemison, E., Kaukinen, C.,

& Knaul, F. M. (2021). Domestic violence during the

COVID-19 pandemic - Evidence from a systematic

review and meta-analysis. Journal of Criminal Justice,

74, 101806.

Roesch, E., Amin, A., Gupta, J., & García-Moreno, C.

(2020). Violence against women during covid-19

pandemic restrictions. In (Vol. 369): British Medical

Journal Publishing Group.

Sifat, R. I. (2020). Sexual violence against women in

Bangladesh during the COVID-19 pandemic. Asian

journal of psychiatry, 54, 102455.

Thaipisutikul, T., Tuarob, S., Pongpaichet, S.,

Amornvatcharapong, A., & Shih, T. K. (2021).

Automated Classification of Criminal and Violent

Activities in Thailand from Online News Articles. Paper

presented at the 2021 13th International Conference on

Knowledge and Smart Technology.

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(3), 760-774.

Wu, C. Y. H., & Loo, B. P. Y. (2017). Changes in novice

motorcyclist safety in Hong Kong after the

probationary driving license scheme. Transportmetrica

A: Transport Science, 13(5), 435-448.

Yang, M., Chen, Z., Zhou, M., Liang, X., & Bai, Z. (2021).

The Impact of COVID-19 on Crime: A Spatial

Temporal Analysis in Chicago. ISPRS International

Journal of Geo-Information, 10(3), 152.

The Impact of COVID-19 on Crime: A Study from the Spatial-temporal Perspective in the Montgomery County, AL

539