Crime in the City


Lab Reports, Maps, Visualization

Introduction and Motivation

In the previous lab on charts and graphs I had come across a choropleth map of crime put out by the New York Police Department(left) which led me to make a map in Tableau Public(right). I was never happy with how the map came out as I could only figure out how to plot the points on the map which quickly became over saturated as there were over 100,000 data entries in the data set. This led me to want to redo the map using a proper GIS application where I can join data to shape files to produce a choropleth map. Furthermore, by bringing in population data, we can compare the a of the counts to one of the rates in order to get a more complete picture.

NYPD Crime Choropleth
Lab 2 Map

Methods

The crime data and police precincts shape files were obtained from New York City’s open data portal and the population data was obtained from the census api. The data was loaded into QGIS where I joined the population data to their census tracts and then aggregated the data based on which police precinct the census tracts were in. I then joined that data and the crime data to the police precinct shape file and used that to calculate the crime rate per 1000 population. From here all that was left was to style the data and add both a legend and a title.

Findings

When looking solely at the crime counts in each precinct(left) we see that the Bronx has a lot of high crime precincts, especially in its southern and central areas, as does Queens and Brooklyn. This is quite different from Manhattan which has a singular hotspot just south of central park near the Empire State Building. With all of this it is quite understandable for people to come to the conclusion that the city is unsafe as many of the precincts but raw counts do not tell the full story. Like with most things pertaining to humans, the more people there are in a region the more crime there will be in that region. As such a better indicator of crime severity is the crime rate as it includes the population of the region in which the crimes occur. By looking at the crime rate within each precinct(right) we see that there only exists one hotspot in New York City and that being the precinct previously mentioned in Manhattan. Now Brooklyn and Queens both have some of the lowest crime rates due to their high population which is a large contrast from them having some of the highest counts. Finally, while the precincts in the Bronx do still have high crime rates, they are not as high relatively speaking as their count counterparts.

Future Studies

There are two main paths that I would like to explore further with this data set. The first is to create maps like this but broken up by the various types of crime to see where they occur. The second is to incorporate one of my findings from the second lab report, that being that crime is not uniformly distributed throughout the week. This would allow for us to if the crime hotspots change throughout the week. Finally, if time permits, I would combine these two paths creating a grid of maps, with day of the week for columns and the type of crime for rows, to show the entire crime trend by type and day.