Crime and Mobility in Baltimore City
Loading...
Links to Files
Permanent Link
Collections
Author/Creator
Author/Creator ORCID
Date
2024
Type of Work
Department
Towson University. Department of Mathematics
Program
Applied Mathematics Laboratory
Citation of Original Publication
Rights
No embargo is requested, catalog data and report are freely available.
Abstract
This study investigates mobility and crime in Baltimore City, where mobility is defined as the number of trips taken to commercial establishments called points of interest (also called POI). The mobility data provides a sample of the number of trips to a POI, the location of the POI, the census block group the POI resides in, and the date range the data is collected for. This study focuses on Part I crimes for the crime data, which includes assault, common assault, shooting, rape, homicide, burglary, robbery, commercial robbery, larceny, larceny from auto, carjacking, auto theft, and arson. These crimes can be classified into two categories, property crimes and violent crimes and property crimes. Violent crimes include aggravated assault, common assault, shooting, rape, and homicide. Property related crimes included burglary, robbery, commercial robbery, larceny, larceny from auto, carjacking, auto theft, and arson. Arson, often classified as a property-related crime, typically revolves around insurance fraud rather than theft, distinguishing it from other property crimes. Part I crime data provides the location of crimes, type of crimes, date/time of crime, and characteristics.
To analyze the relationship between crime and mobility, a quadrat analysis was conducted to determine if crime could be approximated by a piecewise constant intensity via a Poisson-point process model. It is discovered that at no geographical granularity could crime be approximated by a piecewise constant intensity. Through a K-means clustering analysis, it was discovered that the relationship between crime and mobility could be partitioned into two clusters. Clusters are identified as pre- and post-COVID. A χ2 analysis confirms the significance of these clusters, identifying them as pre- and post-COVID. This proves that these clusters must be treated as separate datasets.
Linear regression analysis reveals a positive correlation between certain crime types and mobility, particularly violent crimes. These findings provide insights into the impact of mobility on crime, while considering the effects of the pandemic. A seasonal pattern was found for all crime types, where crimes rose in the warmer months and dropped in the cooler months. The most obvious pattern occurs for violent crimes.