Serious Crimes in NYC:

Is There a Spatial Relationship?

The concentration of crime felony in New York City.

Using NYPD Complaint Data of 2019. The data is filter to felony crimes only. The figure to the right shows the spatial distribution of total felony crimes in New York City which consists of 62 classifications including robberies, dangerous drugs, and dangerous weapons.

Overall, Brooklyn and Bronx can be identified as two the boroughs with the highest crime rates while Staten Island seems to have a relatively low crime rate.

There is a cluster of high crime frequency near the border of Brooklyn and Queens connected to another cluster of lower crime frequency in Brooklyn towards Manhattan. The same pattern observed in Bronx where a high frequency crime neighborhood is surrounded by lower frequency crime neighborhoods.

The crimes in Manhattan are concentrated in midtown and lower parts as well as far uptown. However, none of these neighborhoods has very high crime rates. This doesn’t necessarily mean that Manhattan is safer than Brooklyn. The size of NYC zipcode areas should be taken into account before driving such a conclusion!

Spatially weighted means and spatially weighted standard deviational ellipses are good statistical measures to compare three types of crimes: robberies, dangerous drugs, and dangerous weapons in NYC.

Figure 1. The spatial distribution of (a) dangerous drugs, (b) robberies, and (c) dangerous weapons s in New York City.

Figure 2. Felony crimes distribution in New York City color coded by the type.

The spatial distributions of the three crime types are shown in Fig.1. As expected, the standard deviation ellipses of the three crime types are stretched vertically between Brooklyn and Bronx. This makes sense because high crime rates were found in these boroughs with slight variation from one type of crime to another. In Fig.2 we can see how the spread of the three crime types given the standard deviational ellipse shape and the center of masses is different.

The center of mass for robberies falls closely to dangerous drugs mean. However, the standard deviation ellipse for robberies has a slightly different geometric shape suggesting that the data is less stretched and more concentrated toward the mean center compared to the other two crimes types. This is supported by the choropleth map in Fig.1b where clusters of high robbery rates are mainly concentrated in Bronx and Brooklyn unlike the distribution of dangerous drugs crimes (Fig.1.a) where neighborhoods of high rate are found in all boroughs which explains why the SDE is more stretched.

On the other hand, the center of mass for dangerous weapons crimes is shifted towards Brooklyn. This suggests that dangerous weapons crimes are more concentrated in Brooklyn than Bronx. This is also supported by the spatial distribution in Fig.1c. The overlap of the three SDEs in Fig.2 gives a general idea of the spatial spread of the crimes regardless of their type while the opening figure is more helpful in identifying the troubled neighborhoods.

To find out how we arrived to these results, contact me for the full report.

This exercise was part of GIS & Spatial Analysis Course by Michael Parrott in Columbia University.