Anticipating Customer Churn in Telecommunication Using Machine Learning Algorithms for Customer Retention

Customer attrition is a significant issue and a top priority for large corporations. This study addresses this pressing issue. The dataset is obtained from Kaggle which constitutes of training set and testing set, 80 percent and 20 percent of the entire dataset respectively for identifying customers who tend to unsubscribe in the telecommunications industry. Predicting customer churn with precision is a difficult endeavor, mostly due to the dependence on a singular prediction model in most existing projects. To address this, this study proposes a novel approach that compares multiple prediction models like logistic regression, KNN, random forest, SVV, Gaussian NB, Kernel SVM, Support Vector Machine, to estimate customer churn. With ROC AUC Mean score of 84% for Logistic Regression produced a better result. In the future, the study might be expanded to investigate the evolving behavioral patterns of churn consumers with the application of Artificial Intelligence approaches for predictive and trend analysis to preserve valuable client.