Crime Prediction with Geo Hotspots and Heatmaps using Machine Learning

Abstract

Crime is regarded as the most severe and urgent problem facing our society, and preventing it would be a crucial task. A considerable number of crimes are committed each day. This requires keeping track of all offences and maintaining a database so they may be accessed in the future. However, maintaining a reliable crime record and analyzing this information to help forecast and foundation future crimes is a challenge. This work aims to examine a collection including a selection of crimes and predict the crime that may emerge in the future based on the dispersion of causes. Law enforcement agencies have devised various crime prevention techniques to recognize the gravity of this issue. However, in most situations, the inefficiency and sluggishness of these preventative measures render them incapable of predicting crime patterns for initial prevention. This research provides a regression-based approach incorporating temporal, statistical correlations and other pertinent data to predict the crime patterns of distinct districts concerning states. We employ Auto-Regressive Integrated Moving Average (ARIMA) to evaluate the crime trends of different districts of states for the analysis of occurrences of crimes by employing the crime trends because seasonal information is a valuable addition to applying time-series patterns