Application of an ensemble learning based classifier in crime prediction

Abstract

As an outstanding issue for police, crime prediction has been paid widely attention by researchers. Based on Ensemble Learning method, this paper applies random forest to classifier and introduces a crime prediction method to deeply explore the characteristics of criminal suspects to achieve the purpose of crime prevention. According to calculating attribute importance, the method keeps the important attributes in the algorithm. The reduced attribute set is used to train the random forest model to obtain the crime prediction classifier. The crime data was applied to the proposed classifier which is evaluated by the precision and recall. The experimental results show that the presented classifier is effective.