International Journal of Scientific Research and Engineering Development

Urban safety management has become increasingly complex due to rapid urbanization and evolving crime patterns in metropolitan regions. Accurate forecasting of crime type and occurrence plays a vital role in proactive policing, effective resource allocation, and strategic urban planning. Traditional crime analysis approaches are largely reactive and rely on historical interpretation without predictive capabilities. This paper proposes a Crime Type and Occurrence Prediction System using Machine Learning techniques to forecast daily crime trends and identify high-risk periods. The system utilizes historical crime records, temporal feature engineering, and an ensemble learning approach integrating Random Forest, Gradient Boosting, XGBoost, and Elastic Net models. Advanced temporal features such as lag variables, rolling averages, seasonal indicators, and cyclical encodings are incorporated to capture dynamic crime behavior. A weighted ensemble mechanism enhances predictive performance and reduces model variance. The system is deployed through an interactive Streamlit-based web application, enabling real-time visualization, forecast generation, and analytical insights for stakeholders. Experimental evaluation demonstrates improved accuracy, stability, and reliability in predicting crime occurrences. The proposed system supports proactive crime prevention strategies and data-driven decision-making for law enforcement agencies and urban planners.

Santhosh.G, Dr.K.Banuroopa,"Crime Type and Occurance Prediction Using ML" International Journal of Scientific Research and Engineering Development, V9(2): Page(766-773) Mar-Apr 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.