Customer churn prediction model in enterprises using machine learning

The customer is a vital and integral component of all organizations worldwide. The success of any business hinges on its ability to engage customers effectively in terms of the products or services offered. Research into customer churn is a critical element in the operations of any enterprise. Customer churn prediction (CCP) helps to understand customer interactions with a business and often to identify when customers are likely to cease doing business with the company. In this study, Machine Learning (ML) algorithms are utilized for effective CCP. This study considers various supervised learning models, as the dataset used is labeled. The ML models employed include logistic regression (LR), k-nearest neighbors (k-NN), decision tree (DT), random forest (RF), XGBoost (XGB), light gradient boosting machine (LightGBM), and CatBoost. The ML model is applied to a dataset of customer churn sourced from the Kaggle repository. The results of the experiments are evaluated using several validity metrics, such as accuracy, recall, precision, area under the curve (AUC), and F1-score. The experimental data reveal that LR excels in terms of recall (0.5275) and accuracy (0.581), while the CatBoost model leads in AUC (0.8415), precision (0.6564), and F1-score (0.581). Moreover, LightGBM achieves results close to those of LR and CatBoost. The research findings indicate that the use of ML contributes to the prediction of customer churn. Additionally, the experimental results suggest that LR, CatBoost, and LightGBM outperform other ML models. Utilizing this knowledge, enterprises can develop more effective strategies for customer retention and enhance their business's financial performance.