Crime Forecasting Using Ensemble Learning

Crime analysis plays a vital role in enabling law enforcement agencies to make proactive decisions and enhance public safety. This study analyzes district-level crime trends using a structured machine learning approach involving data preprocessing, clustering, classification, and regression. Preprocessing methods such as standardization, imputation, and one-hot encoding were applied to improve performance. KMeans clustering identified distinct crime patterns, while classification models-Support Vector Machines (SVM), KNearest Neighbours (KNN), Logistic Regression, and Artificial Neural Networks (ANN)-were trained to categorize crimes. An ensemble classifier combining multiple algorithms achieved the highest accuracy of 98.81 percent, outperforming individual models. For crime rate prediction, Random Forest Regression, Linear Regression, and Support Vector Regression (SVR) were used, with the ensemble regressor attaining an R2 score of 0.9458. Results demonstrate the effectiveness of AI-based models in both crime classification and prediction, providing valuable insights for policymaking, efficient resource allocation, and the development of proactive crime prevention strategies.