Developing Predictive Models for Smart Policing Based on Baltimore’s Crime and Product Price Correlation

Inflation is a phenomenon that affects people at different financial levels, especially when it impacts the basic necessities of life, such as food, housing, education, and health care. In situations like these, people find themselves committing acts they would never commit under normal circumstances. In this work, machine learning models are implemented on the data collected from police reports and the criminal incident dataset from 2012 to early 2023, combined with the Consumer Price Index (CPI), to predict a certain crime category. In this paper, data analysis is performed to observe how an increase in product prices affects the frequency of crimes, and popular machine learning models are utilized on the combined crime and CPI datasets to predict a probable crime category. Both multiclass and binary classifications are performed to predict crimes. Accuracy and AUROC scores are derived for the performance evaluation of the models. The proposed methodology will enable law enforcement to take essential steps to reduce crime by initiating crime alerts and dispatching police to deter crimes at probable crime locations.