Prediction of crime tendency of high-risk personnel using C5.0 decision tree empowered by particle swarm optimization

Abstract

The research on the big data in the security and protection industry has been increasingly recognized as the hotspot in case of the rapid development of the big data. This paper mainly focuses on addressing the problem that predicts the criminal tendency of the high-risk personnel based on the recorded behavior data of the high-risk personnel. Therefore, we propose a novel predictive model that is the crime tendency of high-risk personnel using C5.0 based on particle swarm optimization. In this model, the C5.0 decision tree algorithm is first used as a classifier, in which repeated tenfold cross-validation is used and then continuously tuned according to the custom fitness function based on particle swarm optimization. In addition, the classification accuracy, the reduced number of feature subset, specificity and sensitivity under different algorithms are compared. Finally, the proposed model has higher accuracy through the optimal value of the particle position, the error rate of the cost under different iterations and the trend and the concavity and convexity of ROC curve. The experimental results show that the proposed model has a good effect on the predictive classification, which may provide guidance for predicting crime tendency of high-risk personnel.

Keywords

• big data
• high-risk personnel
• particle swarm optimization
• c5.0
• k-fold cross-validation