Github: PatrickGitundu
NYUID: pgm293
Problem Description:
In public safety, response delays (even to the minute) to crimes as they happen could mean the difference between life and death. Despite a decline in the number of “crime in progress” calls over the last three years, NYPD response times to crimes in progress and non-critical calls are all up, according to the 2016 Mayor’s Management Report released in September of this year. To get a grasp of the dynamics in response times could be invaluable. The NYPD releases a data set of the major felonies committed in the city per precinct as well as the response times for those crimes.
Goals:
By analyzing this data set of crimes by precinct and response time, can we :-
1. Predict future response times by the precinct and crime. What will be the response time to crimes in the future?
2. Identify trends, events or periodicity with regards to the response time?
3. Identify the ‘slowest’ precincts?
4. Spatially visualize the response times for crimes in progress using a ‘responsiveness’ scale for each precinct (zip code might be possible as well)?
Data:
The data to be used will be the NYPD public Indicators dataset hosted on New York City open data platform. The dataset contains response times for crimes in progress. The crimes recorded are split according to the 7 major felonies as well as by precinct.
Analysis:
I will look to predict future response times using regression. I will also carry out several time-series analyses with the data. I will look for trends by smoothing, look for any events through thresholds, lastly I will search for periodicity using ARMA. In addition to the above I will also look to use spatial autocorrelation through LISA to create a spatial ‘heat map’ of the average response times as seen geographically by precinct service area boundaries and zip code. This heat map will show the responsiveness of precincts on a scale from 1 to 10.
Deliverable:
As a deliverable, I would like to come up with a report showing my statistical and spatial conclusions about the response times and crimes as handled by precinct and time. This could be useful in identifying slower precincts and helping the city administration to better allocate resources towards bringing the response time to crimes in progress down.
References:
Anthony Shorris, Mindy Tarlow. The Mayor’s Management Report – September 2016.