This research delves into using various machine learning models, such as Artificial Neural Networks, Decision Trees, Support Vector Machines, Random Forests, Logistic Regression, and advanced gradient boosting techniques including XGBoost, LightGBM, and CatBoost. The study focuses on predicting customer churn in the telecommunications sector, utilizing a publicly accessible dataset. To address the complexities of imbalanced datasets, the team employed innovative data sampling strategies like SMOTE, combined with Tomek Links and Edited Nearest Neighbors. Emphasizing precision, recall, F1-score, and ROC AUC metrics, the paper significantly enhances the understanding and performance of models in churn prediction, particularly following the application of hyperparameter optimization techniques.

https://www.mdpi.com/2227-7080/11/6/167