Applying AI-Driven Predictive Policing: A Machine Learning Approach to Crime Prediction and Prevention

Abstract

This study is intended to explore the use of machine learning models in predictive policing by using various methods to improve crime forecasting and prevention. Several combinations, such as the random forest, support vector machine (SVM), and neural network models, were used to predict the crime hotspot and its rate. To train and validate the models fed with a dataset of crime reports and demographic data, our outcomes proved that they predicted crime hotspots and rates sufficiently and quite well; random forest stood out as the best among them. These findings will inevitably have important implications for policing and public safety. If predictive policing is implemented effectively, as demonstrated in this model, crime and community safety will be improved. The paper contributes to the rising interest in predictive policing and differences in machine learning applications in crime prediction.