CUSTOMER CHURN PREDICTION USING LOGISTIC REGRESSION AND DECISION TREE (CART) TECHNIQUES

Abstract

The term “customer churn” is used to indicate those customers who are about to leave for a competitor or end their subscription. Customer churn or customer attrition has become an important issue for organizations particularly in subscription-based businesses, where customers have a contractual relationship which must be ended. According to numerous studies, acquiring new clients is significantly more expensive than keeping the ones that already have. As a result, businesses are concentrating on creating precise and trustworthy predictive models to pinpoint potential clients who will churn in the near future. This model uses data from telecom companies on a range of aspects, including customers who left within the last month, services that each client has signed up for, demographic data about customers, and customer account information. The model is presented using machine learning techniques, particularly Logistic Regression (LR) and Decision Tree (DT), to forecast churn for telecoms companies. Comparisons are made to determine the algorithm's efficacy using the provided dataset. The results from a strategy based on Logistic Regression (LR) can predict the telecom market better than Decision Tree (DT) techniques.