Customer Churn Prediction in Telecommunication Industry. A Data Analysis Techniques Approach

Telecommunications is one of the most dynamic sectors in the market, where the customer base is an important pawn in receive safe revenues, so is important to focus attention is paid to maintaining them with sun active status. Migrating customers from one network to another varies among telecommunication companies depending on different factors such as call quality, pricing plan, minute consumption, data, sms facilities, customer billing issues, etc. Determining an effective predictive model helps detect early warning signals when churn occurs and assigns to each customer a score called chum score that indicates the likelihood that the individual might migrate to another network over a predefined time period. To this extent, the present paper uses more than 10k customers sample of a telecommunication company and tries to analyse the churn behavior. The aim of the paper is both to test the efficiency and performance of the most commonly used data mining techniques to predict the chum behavior and to underline the main indicators that can be used when conducting such analyses. knowing the magnitude of the chum phenomenon, the company can prevent the instability that is going to occur by applying a series of measure in order to increase the retention of the current customers.

Automated summary

This paper examines customer churn behavior in the telecommunications industry and tests the effectiveness and performance of commonly used data mining techniques. By determining predictive models and key indicators, early warning signals of customer churn can be detected, and measures can be taken to increase customer retention.