Developing Machine Learning Based Predictive Models for Smart Policing

Abstract

Crimes are problematic where normal social issues are confronted and influence personal satisfaction, financial development, and quality-of-life of a region. There has been a surge in the crime rate over the past couple of years. To reduce the offense rate, law enforcement needs to embrace innovative preventive technological measures. Accurate crime forecasts help to decrease the crime rate. However, predicting criminal activities is difficult due to the high complexity associated with modeling numerous intricate elements. In this work, we employ statistical analysis methods and machine learning models for predicting different types of crimes in New York City, based on 2018 crime datasets. We combine weather, and its temporal attributes like cloud cover, lighting and time of day to identify relevance to crime data. We note that weatherrelated attributes play a negligible role in crime forecasting. We have evaluated the various performance metrics of crime prediction, with and without the consideration of weather datasets, on different types of crime committed. Our proposed methodology will enable law enforcement to make effective decisions on appropriate resource allocation, including backup officers related to crime type and location