A REVIEW ON CUSTOMER CHURN PREDICTION DATA MINING MODELING TECHNIQUES

Objectives: To find one of the best data mining techniques in telecommunication especially in customer churn prediction. Methods/Statistical Analysis: This paper presents a review of customer’s churn prediction in the telecommunication. The study shows a large number of attributes that are used to put into practice to develop customer churn prediction model by the large number of reviewer. These attributes are segmentation, account info, billing info, call dialup types, line-info, and payment info, and complain info, service provider info, and services info. In this study appropriate modeling techniques such as LR, NNM, DT, FL, CMC, SVM and DME are discussed for the churning purpose. Findings: The Review shows that to find customer churn prediction depends on the objectives of decision maker’s e.g. DT and SVM with a low ratio used, if interested in the true churn rate and false churn rate. The Logistic Regressions might be used if looking for the churn probability. DMEL modeling technique is impractical and ineffective for churn prediction on a large dataset with high dimension. Application/Improvements: The Technique proposed in this paper will overcome discussed issues and it will be applied on those customers who want to leave in near future and predict them based on some parameters.  Keywords: Customer Prediction, Churn Prediction, Data Mining, Data Modeling