Original research

PREDICTING CUSTOMER CHURN WITH ADVANCED MACHINE LEARNING APPROACH

Pages 93-102

10.61552/JAI.2026.02.004

Sachin Gaur, Dhruv Lingwal

• View Article
• PDF 1123 K

Abstract Nowadays telecom business is transforming into a competitive market to take edge over their rivals. The customers are key element for any form of business, so it affects growth, profit, revenue, workload of the particular firm. As a result, large amount of data is collected and updated on regular basis by the customer feedback and behavior analysis of customer. As developing a new lot of customers are more expensive than retaining the existing customers. Telecom business analysts give their verdict about the existing customers likely to churn on the basis of customer behavior and response. Now customer retention policy which is initiated by various telecom businesses for prediction of customer churn allows firms to yield appropriate measures. For performing such measures, firms are applying different techniques and methods to distinguish their customers early through customer retention strategies. Machine learning techniques are utilized for customer churn prediction and classification, the main of aim of the model is to maintain the responses of existing customers’ through the help of various machine learning algorithms. In this research, variety of model is developed for customer churn prediction such as Rule Based System (RBS), Attention Based Model (ABM) and Explainable AI (XAI). These set of models are utilized to analyze churn data gathered from customers, which results in effective predictions of churn customers making business management efficient and make proactive decisions during their churn period in order to avoid loss of customer churn as well as profit. The proposed system achieves the AUROC score of 84.8% for RBS, 85.6% for ABM and 88.4% for XAI. Moreover, the research work improves customer churn prediction along with additional features and has potential to be effectively equipped with telecom business.

Keywords: Customer Churn, Rule-Based System (RBS), Attention-Based Model (ABM), Explainable Artificial Intelligence (XAI), random forest, bagging.

Recieved: 18.04.2025. Revised: 19.06.2025. Accepted: 06.07.2025.