Application of data mining for behavior pattern recognition in telecommunication

Abstract

In telecom industry, mobile subscribers produce data traffic while online every day. These data traffic suggests that certain characteristics of the behavior. The application of data mining helped to analyze and identify the features from the data traffic. In this paper, we use exponential binning of data preprocessing technology to smooth the data sets and keep reduce the noise. By using K-means algorithm to cluster the data traffic stream, we aim to mining subscribers’ behavior characteristics from clusters, provide support for churn prediction, target marketing, and fraud detection.