Enhanced Crime Hotspot Prediction and Visualization for Women’s Safety Through Deep Learning

Crime hotspots are geographic areas with elevated levels of criminal activities compared to other regions. Women in these hotspots are at increased risk of experiencing various forms of criminal behavior, including sexual harassment, assault, domestic violence, stalking, and human trafficking. Identifying these hotspots is crucial for effective crime prevention and resource allocation by law enforcement agencies. This project presents Safety Locator, a predictive system that uses multimodal deep learning to identify and map crime hotspots where women are particularly vulnerable. The system utilizes the Deep Explainable Decision Tree model, a machine learning algorithm that analyzes historical crime data to predict the likelihood of criminal activity in specific areas.  The Deep Explainable Decision Tree model classifies data into crime hotspot categories and visualizes these areas on Google Maps. By analyzing factors such as the number and type of reported crimes, the timing of incidents, and crime locations, the system generates a detailed map highlighting high-risk areas. This map can be shared with the public to increase awareness and promote safety. The development process includes data pre-processing, feature selection, model training, evaluation, hyperparameter tuning, and prediction. This model’s performance is assessed using metrics like accuracy, precision, recall, and F1-score, with hyperparameters optimized through cross-validation. Safety Locator aims to assist law enforcement agencies in preventing crime and enhancing public safety by pinpointing areas with high crime probabilities. By creating safer environments for women, the system supports efforts toward gender equality and improved quality of life.