Outlier analysis in data mining is to find the deviation and abnormal data in the databases. By finding and explaining the outliers, we can apply them to detecting the business frauds effectively. This paper analyzes the common characteristics of fraudulent behavior of customers in telecom industry systematically. Based on the outlier-finding by clustering in data mining, we propose an effective solution to forecast the customers who are maliciously in arrears. Coupled with the actual application of forecasting the customers who are maliciously in arrears in telecom industry, we propose the specific method to forecast this kind of customers by using Kohonen neural network clustering algorithm.
Wu, Sen; Kang, Naidong; and Yang, Liu
"Fraudulent Behavior Forecast in Telecom Industry Based on Data Mining Technology,"
Communications of the IIMA: Vol. 7
, Article 1.
Available at: http://scholarworks.lib.csusb.edu/ciima/vol7/iss4/1