Spatial Clustering and Machine Learning for Crime Prediction: A Case Study on Women Safety in Bhopal

Yamini Sahu,Vaibhav Kumar; Proceedings of the Asian Conference on Computer Vision (ACCV) Workshops, 2024, pp. 648-663

A crime is an unlawful action subject to punishment by a governing authority, causing harm not only to individuals but also to the well-being of a community, society, or the state. According to the latest annual report from the National Crime Records Bureau of India, there were 445,256 cases of crimes against women registered in 2022, representing a 4% increase from the 428,278 cases reported in 2021. India experiences an alarming rate of 51 cases of crimes against women per hour. These figures underscore the urgent need for proactive measures to ensure women safety and secure their continued contribution to the country's development. Our research employs a predictive approach to address crimes against women, utilizing spatial analysis and crime prediction models to pinpoint high-risk areas in urban settings and accurately forecast crime trends. We focused on Bhopal, the capital of Madhya Pradesh, one of India's 28 states, and gathered crime data against women from 30 police stations in Bhopal. Our study showcases the application of spatial clustering techniques to identify hotspots for various crimes against women (murder, attempted murder, rape, gang rape, kidnapping, dowry-related offenses, and molestation). Additionally, we employed machine learning regression models and advanced forecasting techniques to predict crime rates. Our model, based on decision tree regression, exhibited a very low mean squared error of 0.0417 and a mean absolute error of 0.083. Furthermore, our analysis revealed that classical machine learning regression models outperformed advanced forecasting models, such as long shortterm memory, given our limited dataset. Thus, detecting and predicting crime hotspots derived from historical crime data can enable law enforcement agencies to develop targeted intervention strategies customized for specific crimes occurring at particular locations.