Using data mining techniques to predict customer churn rates in telecommunications companies: A case study on NBTel in Nineveh Governorate

Abstract

This study addresses customer retention challenges in the telecommunications sector by applying data mining and machine learning techniques to predict customer churn at NBTel in Nineveh, Iraq. A dataset of 25 customers was analyzed across six dimensions: demographic information, usage behavior, subscription and contract details, complaints and technical support records, marketing campaign interactions, and financial data. Statistical analyses were conducted using SPSS, including descriptive statistics, ANOVA, Chi-square tests, correlation analysis, logistic regression, decision tree, discriminant analysis, and cluster analysis. Results indicated that 40% of customers were at high risk of churn, 24% at moderate risk, and 36% at low risk. Key determinants of churn included activity level, contract type, number of complaints, payment delays, and satisfaction with technical support. The proposed predictive model achieved an overall classification accuracy of 84% and provides actionable recommendations for targeted retention strategies.