An Efficient Technique for Feature Selection to Predict Customer Churn in telecom industry

Abstract

The evolution of technology has a great impact on the telecom industry, which has grown rapidly from telegraph to present high speed network. This rapid growth has resulted in the establishment of many telecom sectors which in turn has given rise to a stiff competition among them. Telecom sectors with improved technology needs to handle the large set of subscribed customer base. Now a days, in addition to acquisition of new customers to increase the company revenue, retaining the old customers is also found to be of much importance. So, all the telecom industries are concentrating on building a best predictive model in order to determine the churn rate. In this paper we mainly concentrate on refining the telecom dataset by applying the Pre-processing, feature selection and feature extraction techniques. The refined dataset is created to provide the prediction accuracy similar to or greater than the original dataset with less computation.