Forecasting Criminal Activity: an Empirical Approach for Crime Rate Prediction

Abstract

A crime is an act that is prohibited by law and is punishable by a fine, imprisonment, or other legal action. Every day, the news of criminal activity fills our news channels and social media platforms, portraying an image of a society in which crime is an ever-present concern. Crime rate prediction using machine learning is a critical undertaking in today’s society to handle the ever-present concern of criminal activities. This paper presents a comprehensive approach to predicting crime rates by leveraging machine learning algorithms and data analysis techniques. The proposed system utilizes historical crime data to develop predictive models that identify high-risk areas and potential future crime trends. The dataset, sourced from the National Crime Rate Bureau (NCRB), undergoes preprocessing, which includes feature engineering and data augmentation. Five distinct models—nearest neighbor, support vector machine, random forest, decision tree, and XGBoost—are assessed for their predictive performance. Among these, the XGBoost Regressor exhibits the highest accuracy in predicting crime rates for 8 distinct crime categories in 19 Indian metropolitan cities. The results indicate a promising accuracy of 93.20% using the selected model, showcasing the potential of machine learning in crime prediction. By targeting resources toward high-risk regions, law enforcement authorities may successfully suppress criminal activity and promote community safety. This study highlights the significance of predictive policing and its role in designing future crime measures. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.