Prediction of Criminal Tendency of High-risk Personnel Based on Combination of Principal Component Analysis and Support Vector Machine
Fang Ou Yang, Chunxue Wu, Naixue N. Xiong, Yan Wu · International Journal of Social Relevance & Concern · 2018
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.Since reducing the dimension of data will improve the accuracy of the classification directly, we propose a predictive model that combines Principal Component Analysis (PCA) and Support Vector Machine (SVM) to predict the criminal tendency of high-risk personnel.In the model, at first, PCA is used to reduce the dimension of data preprocessed.And then SVM with different kernel functions will predict whether the high-risk personnel have a criminal tendency or not, moreover, the loss penalty parameter can be obtained by the k-fold crossvalidation which could ensure the minimum prediction error.At last, theoretical analysis and simulation results prove that the accuracy could vary with the different kernel function in SVM.In addition, through the evaluation of Receiver Operating Characteristic (ROC) curve, the calibration diagram and the lift chart, the SVM under the radial basis kernel function outperforms the other kernel functions in terms of prediction accuracy.