A Novel Customer Churn Prediction Method Based on Competitions in the Telecommunication Industry

摘要

In this paper, we bring forward a novel Competition-Based Churn Prediction (CBCP) algorithm. Instead of studying customer's behavior as other researches did, we think that most important fact of customer churn is from the competition of other companies. Based on that fact, for each current customer, CBCP calculates the cost difference between self company and competitor companies. Then predict whether the customer will churn. The experimental result shows that CBCP can accurately predict the churn behavior of customers. Based on the information of customer consumption behavior obtained from CBCP, a Customer Keeping Strategy Model (CKSM) prototype is brought out to try to keep the customers that have higher churn possibility.