Crime information
This dataset contains a variety of details about crime at the incident level, including location, from 2006 to 2016. This will allow me to aggregate crimes by type and location. The difficult part will be justifying if 2006 crime data can be linked to 2005 tree census data. I will make a strong assumption that the one year difference is negligible.
Additional information
This site contains demographic, social, and economic data for New York City as collected on the American Community Survey. I plan to control for variables that are known to be predictive of crime rates such as population size and economic conditions by including them my regression model.

Analysis

Given that the primary research question seeks to characterize the relationship between two measures, I plan to use regression techniques to analyze the data. Specifically, plan to model crime rate within PUMAs using multivariate regression, including covariates such as the number of street trees per PUMA and apparent level of care for tree ('number of signs of stewardship', 'tree appears to be dead').

References

Researchers have explored the relationship between street trees and crime rates in the past. In 2012, Troy, Morgan Grove,  and O’Neil-Dunne \cite{TROY2012262}examined the relationship between street trees and crime in Baltimore and found an inverse relationship between tree canopy cover and crime incidence. Similar investigations have taken place in other cities such as New Haven, Portland, and Chicago.