Real Time Alert System based on Crime Area Mapping

The urban landscape of Pune, India, ensuring public safety is an ongoing challenge. This project introduces a novel real-time alert system founded on the comprehensive analysis of historical crime data . To overcome data acquisition hurdles, the proposed system used meticulously extracted incident details from thousands of FIRs, translating Marathi text, and structured it into a usable format. Leveraging geocoding techniques, the proposed system Kernel Density Estimation (KDE) to spatially predict safety probabilities, complementing our safety-centric approach. Survey-based target variables augmented the dataset, enabling the development of a machine learning model. Linear regression emerged as the most effective model, shedding light on factors influencing safety scores, including proximity to police stations and highways. This model is integrated with Android application designed to enhance user safety by combining SOS alert and emergency calling features and safety score prediction.