EUVIMEDEUVIMED entdecken
Crossref

Customer Churn Prediction using XGBoost

Sayali Tanpure, Pooja Patil

International Scientific Journal of Engineering & Management · 24.03.2026 · 05 (03) · 1-9

DOI: 10.55041/isjem05863 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Scientific Journal of Engineering & Management, online erschienen am 24.03.2026. Verfasst von Sayali Tanpure und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Customer churn prediction is a critical task for businesses aiming to improve customer retention and maximize long-term profitability. This study proposes an advanced churn prediction model using XGBoost, an optimized gradient boosting technique designed for high performance on structured data. The model is evaluated using multiple performance metrics, including accuracy, precision, recall, F1-score, and AUC-ROC. Experimental results demonstrate that XGBoost significantly improves recall and AUC-ROC compared to traditional Gradient Boosted Decision Trees (GBDT), making it more effective in identifying potential churn customers. The findings highlight the importance of using advanced ensemble methods for predictive analytics and provide actionable insights for developing targeted customer retention strategies.Key Words: Predictive Analystics, Customer Retention, Ensemble Learning

Zitieren

APA

Tanpure, S., & Patil, P. (2026). Customer Churn Prediction using XGBoost. International Scientific Journal of Engineering & Management, 05(03), 1-9. https://doi.org/10.55041/isjem05863

Vancouver

Tanpure S, Patil P. Customer Churn Prediction using XGBoost. International Scientific Journal of Engineering & Management. 2026;05(03):1-9. doi:10.55041/isjem05863

BibTeX
@article{Tanpure2026,
  author = {Tanpure, Sayali and Patil, Pooja},
  title = {Customer Churn Prediction using XGBoost},
  journal = {International Scientific Journal of Engineering & Management},
  year = {2026},
  volume = {05},
  number = {03},
  pages = {1--9},
  doi = {10.55041/isjem05863}
}
Kostenlos starten →