SVM-based Telecom Customer Fraud Detection Technology Research

In recent years, with the rapid development of the telecom market, the telecommunications industry customers in the fraud also in increasing. In order to prevent and detection of fraud occurs, solve customer relations management of the telecom operators, telecom customers fraud detection system using pattern recognition, data mining tools such as telecom customers to the consumption of the specific analytical process, to customers in classification, combined with the telecom customer behavior, and on the basis of telecom customers complete fraud detection modeling, fraud detection, so as to reduce the harm of telecom customers fraud, reduce the telecommunication operation risk.This paper expounds the telecom customers the damage caused by fraud and telecom customers fraud system research and development of the importance and urgency; Introduced the research and development telecom customers fraud system involved such as machine learning, statistical learning theory and related theory knowledge; Based on the single vector support vector machine was first classification, using support vector machines to secondary classification, based on single vector support vector machine, used for telecom customers fraud detection model classification method; Analyzes the telecom customers fraud detection modeling and validation process.Research shows that:the telecom customers based on SVM fraud detection technology application research, has broad market prospects; Detection model can avoid the risk of potential intelligent behavior to the customer to identify fraud, and then to telecom customers in fraud small scale, the conduct of the large dispersion test.