Spatial Analysis: Predictive Policing on Narcotics
The public sector has grown reliance on machine learning algorithms in forecasting where crimes will happen and thereby allocating police officers strategically in the right time and space to prevent crime. Though the intentions of these “predictive policing” tools are to reduce cost and protect public safety, pathologies in data collection and machine learning models can lead to very biased outcomes. This analysis will examine the accuracy and generalizability of our predictive policing model on Cannabis Possession crime in Chicago.
disclamer: this project is a part of a graduate course