Fraud Analysis Of Telecom Customer Based On Bayesian Classification

With the expanding of Chinese telecom market, the fraud phenomena has been increasingly become a serious problem, which not only causes the financial loss, but also restrains the operation from farther development Confronted with the increasing serious fraud phenomena, telecom enterprise must take some methods to defend them. At present, using the classification method, one of data mining technologies, to analyze and predict the customers' behavior is effective against the fraud in telecom industry.The thesis of this paper is the anti-fraud technology in telecom industry. Based on the comparing among many methods, a new method of using Naive Bayesian classification to prevent the fraud in telecom was presented. Meanwhile, the solving method using the Naive Bayesian classification was put forward and the classification model was designed and implemented. The main jobs here include:1. Firstly, the research background and the research meaning were analyzed, then the current research of anti-fraud technology in telecom industry was analyzed and the main anti-fraud ways were discussed.2. The Bayesian classification technology in Data mining was discussed, and the research was emphasized on the basic principle and the work procedure of the Naive Bayesian classification technology.3. The general procedure of handling fraud in telecom industry using the Naive Bayesian classification was presented and an example was given to illustrate its specific idea.4. The solving system of using the naive Bayesian classification to prevent the fraud behavior was analyzed and designed.5. The anti-fraud classification model was implemented and evaluated by testing data.