A Comparative Study of XGBoost, LightGBM, and CatBoost Models for Customer Churn Prediction in the Banking Industry
Jurnal Pepadun · 15.08.2025 · 6 (2) · 178-187
DOI: 10.23960/pepadun.v6i2.277 ↗Kurz gefasst
Fachartikel in der Zeitschrift Jurnal Pepadun, online erschienen am 15.08.2025. Verfasst von Rivaldo Rivaldo und 3 weiteren Autor:innen. Die Arbeit wurde bisher 6-mal zitiert (Crossref). Der Volltext ist frei zugänglich (Open Access, CC BY-NC-SA 4.0).
Abstract (Originalsprache)
Customer churn is a critical issue in the banking industry, as retaining existing customers is more cost-effective than acquiring new ones. High churn rates can negatively affect profitability and long-term business sustainability, making churn prediction a key focus in customer relationship management. With the rise of digital banking and the availability of large-scale customer data, machine learning techniques have become valuable tools for identifying at-risk customers. In particular, gradient boosting algorithms have shown promising results in classification tasks involving structured data. This study compares the performance of three ensemble machine learning models XGBoost, LightGBM, and CatBoost in classifying churn using a publicly available banking customer dataset consisting of 10,127 records and 23 features. The evaluation is conducted using two data-splitting schemes (80:10:10 and 70:15:15), and four performance metrics: accuracy, precision, recall, and F1-score. The results indicate that XGBoost achieved the highest overall performance (98.3% accuracy in split 1 and 96.4% in split 2). LightGBM demonstrated competitive accuracy with significantly faster training time, while CatBoost offered strong predictive capability but required longer computation. These findings suggest that model selection in churn prediction depends on the trade-off between predictive performance and computational efficiency.
Zitieren
APA
Rivaldo, R., Taufik, R., Ilman, I. S., & Wulansari, O. D. E. (2025). A Comparative Study of XGBoost, LightGBM, and CatBoost Models for Customer Churn Prediction in the Banking Industry. Jurnal Pepadun, 6(2), 178-187. https://doi.org/10.23960/pepadun.v6i2.277
Vancouver
Rivaldo R, Taufik R, Ilman IS, Wulansari ODE. A Comparative Study of XGBoost, LightGBM, and CatBoost Models for Customer Churn Prediction in the Banking Industry. Jurnal Pepadun. 2025;6(2):178-187. doi:10.23960/pepadun.v6i2.277
BibTeX
@article{Rivaldo2025,
author = {Rivaldo, Rivaldo and Taufik, Rahman and Ilman, Igit Sabda and Wulansari, Ossy Dwi Endah},
title = {A Comparative Study of XGBoost, LightGBM, and CatBoost Models for Customer Churn Prediction in the Banking Industry},
journal = {Jurnal Pepadun},
year = {2025},
volume = {6},
number = {2},
pages = {178--187},
doi = {10.23960/pepadun.v6i2.277}
}
EUVIMED