Abstract

This paper attempts to predict property crime with statistical theory as well as Python machine learning algorithm and report the difference. Property crimes include charge codes related to burglary or breaking entry, theft and unauthorized use. The methods adopted include classification, clustering, pattern detection and interactive visualization. The experimentation is conducted on a dataset containing the city of Cincinnati crime records from 2010 - 2019. Result comparison of the two predictive models will help correlate factors which might help understand the future scope of crimes and contribute to an effective and efficient predictive policing program within the city of Cincinnati.