A Data Science Approach to Analysing and Understanding Crime Patterns for Improved Law Enforcement Strategies

Abstract

This project focuses on analyzing crime patterns across different states and union territories in India using historical crime data. The dataset contains information on various crime types recorded annually, enabling temporal and spatial analysis. The study applies machine learning techniques, including K-Means clustering, to group states with similar crime trends. Classification algorithms such as Random Forest, Logistic Regression, and Support Vector Machines (SVM) are used to predict crime categories based on historical patterns. K-Means clustering helps in identifying regions with high, medium, and low crime incidence. Random Forest provides insights into the most significant factors influencing crime rates. Logistic Regression is employed for probabilistic prediction of crime occurrence. SVM is used to classify crime types with optimized decision boundaries. The integration of clustering and classification offers both descriptive and predictive analysis of crime patterns. Overall, the system supports policymakers and law enforcement agencies in understanding trends, allocating resources effectively, and implementing targeted crime prevention strategies.