A HYBRID MACHINE LEARNING MODEL FOR CRIME RATE PREDICTION

Crime prediction is vital for public safety and resource management. This study developed a hybrid machine learning model integrating Convolutional Neural Networks (CNN) and K-means clustering for crime rate prediction. Historical crime data from Mubi and Yola from the year 2015 to 2023 yielded training and testing accuracies exceeding 90%, surpassing traditional models (Random Forest and Decision Tree Classifiers). Results underscore the effectiveness of CNN and K-means integration in recognizing spatial patterns and clustering data, demonstrating improved predictive accuracy and forecasting capabilities of predicting crimes up to 2030. This research contributes to advanced crime prediction systems, informing law enforcement agencies' proactive crime prevention and resource allocation.