Analyzing criminal activity is an extremely challenging task. The sheer volume of incidents make it a time-consuming and man-power intensive task for police departments to detect any patterns of crime manually.
In this thesis, we compare two point-process based models --- the Piecewise-Constant Conditional Intensity Model and Multiplicative Forests --- in their ability to analyze crime. Specifically, we compare the performance of both models in learning temporal dependencies in crime incidents. To accomplish this, we make use of 15 years of crime data provided by the City of Chicago. Additionally, we also look at the seasonality of crime, the types of crime most indicative of other crimes, and other consistently-recurring themes. We see promising results that could help authorities with predictive policing and better tackling criminal activity.