An Enhanced Framework for Intelligent Crime Pattern Recognition

Abstract

Crime pattern recognition is one of the most significant activities of the common of the intelligent and LEO (Law Enforcement Organizations) all over the world. Usually they collect national and international crime related record to prevent upcoming attacks and use a limited number of law enforcement resources in an optimum manner. Now a day's, crime ratio is increasing at a higher rate. It is a tough to predict the happenings of crime due to many different elements. It is difficult to predict like which criminal would do the crime but the place and the possibility of the incident of high crime rate in specific time of day and high crime rate in month of the year can very well be predict. And also the most occurrence of type of crime can be predict. The system developed will help to speed up the method of solving crimes for the law enforcement agencies. Data mining is a very good tool that can be used efficiently for studying huge databases and extract important analytical results. In this work, Crime data for the last 15 years is analyzed using two different data-processing approaches. Two predictive models, Random Forest and boosted decision tree, are implemented and a crime prediction accuracy obtained 96% when predicting crime.