A Data Mining Approach for Analysis of Telco Customer Churn

Telco is a telephone service provider that gives data communications access. Information technology and the capability to stay connected and communicate all day are becoming the modern living trend. Telco company offers different promotions and services to attract consumers to select their telco company. The current service is inexpensive, and the quality of most telco companies is the same. Understanding the customer churn rate is crucial as it is highly associated with customer retention. This paper classified customer churn based on various attributes and utilized four different machine learning algorithms: Averaged Perceptron (AP), Bayes Point Machine (BPM), Decision Forest (DF), and Logistic Regression (LR). Microsoft Machine Learning Studio is used for Implementing and Testing Classification Models. The models are evaluated on Telco Customer Churn (TCC) dataset in terms of accuracy, precision, recall, and error. The experimental results show that the LR model has an average accuracy of 80.1% compared to the other three models, in which AP scores 79.92%, BPM scores 80.04%, and DF scores 77.04%. The obtained results can provide significant insights for telecommunications companies to recognize market demands and develop appropriate strategies to enhance customer retention.