Time space cluster analysis of crimes in Chicago and New York
In this exercise, I analyzed crime trends in the city of Chicago, Illinois, and related that information to the city’s neighborhoods, or census block groups. Specifically, I identified temporal patterns of drug-related crimes that resulted in arrests over the last 10 years (from 2012 to 2022) and compared them with the income levels of each census block group from the year 2019. I left cannabis-related arrests out of the analysis, provided that it has been legalized for recreational use in the state of Illinois since 2020. Clearly, this omission does not paint a complete picture since cannabis was illegal during the majority of the chosen time frame, i.e., between 2012 and 2020. However, since I wanted to include drug-related arrests leading up to the present, I decided it was best to leave it out of the analysis.
Procedurally, the analysis comprised of:
Creating a space-time cube by aggregating point data (crime incidents over time).
Determining the spatial autocorrelation of both crime and income to find out the intensity of spatial clustering at various distances.
Running an emerging hot spot analysis (Getis-Ord Gi*) using the space-time cube to determine trends in the clustering of the data through time.
Running a “regular” hot spot analysis (also Getis-Ord Gi*) on the income variable to determine where clusters of high and low incomes are likely to be.
Union the results of both hot spot analyses.
The results can be seen below.
What the final result shows (to the right) is a bivariate choropleth map where the increasing saturation of green colors represents low to high income and blue colors represent low to high drug-related arrests throughout Chicago. It is clear that there is a circular hot spot in the western part of the city that stands out from the rest. The majority of that area seems to be of lower income, but maybe not to a degree that necessarily merits further investigation. Perhaps the more interesting part of the result is the northeastern part of the circular hot spot area, where there is a hot spot of both high drug-related arrests and high income. However, this does not necessarily mean that people with high income are arrested more frequently in census block groups with hot spots of high income, and could simply be because of an overlap between neighboring but overlapping areas with different levels of affluence.
This exercise focused on analyzing crime trends in New York City. The premise was that since the city has experienced a reduction in major crimes like murder, robbery, and burglary over time, more intricate patterns may become apparent if analyzed spatially. Specifically, a potential relationship between public housing, which offers subsidized housing to low-income residents, and major crime was considered. These housing complexes are scattered throughout the city but are concentrated in specific areas of The Bronx, Brooklyn, and Manhattan.
The exercise aimed to:
Perform an emerging hot spot analysis of felony assault incidents between 2010 and 2019.
Explore how these spatial and temporal patterns relate to the distribution of public housing.
These aims sought to determine if crime hot spots are more prevalent near public housing areas and if there are differences among the five boroughs in terms of crime patterns and their connection to public housing locations.
The maps below show the result of this analysis.
The result indicates that there is an overwhelmingly ‘Persistent Hot Spot’ of crimes that characterizes the chosen time frame in a 500-meter proximity of public housing projects in New York City. Looking at Manhattan in the maps above, the clear and consistent ‘Persistent Hot Spot’ pattern may potentially be explained by the higher cluster density of public housing there. Therefore, if it is true that there exists a general relationship between socio-economic conditions, such as levels of wealth, and the prevalence of crime for people living in public housing projects, then this result may serve as evidence for that. However, it is also important to note that the majority of Manhattan is not comprised of public housing developments, yet is almost entirely categorized as a persistent hot spot on the map on the left. In other words, this underlines the importance of being careful about making generalizations and jumping to conclusions based on the data chosen given that so many other factors could play a part in the outcome of this sort of analysis.