AI Based Crime Pattern Prediction

Abstract

The increased rate of urban crime raises important challenges to the law enforcement agencies, and there is a special place reserved for rapidly growing cities like Bengaluru. This paper presents an Artificial Intelligence-driven predictive crime analysis and safety recommendation system designed specifically for Bengaluru city. By using machine learning models such as Random Forest and XGBoost, the proposed system predictively determines the prospective crime types based on spatio-temporal and situational factors. The XGBoost model, after exhaustive data cleaning and feature engineering, produced an accuracy of 93.1%, surpassing other classifiers. It offers real-time safety insights, visualizes crime trends, and provides an interactive heatmap for situational awareness. This research demonstrates the feasibility of deploying AI-powered predictive policing systems at a city level to enhance urban safety and proactive response planning.