Prediction of Telecommunications Customer Churn Based on Hybrid Machine Learning and Deep Learning Algorithms

Tejada-Vicente L.; Rosado-Oliden D.; Mauricio-Santos D. · Proceedings of the 2024 IEEE 31st International Conference on Electronics, Electrical Engineering and Computing, INTERCON 2024 (2024) · DOI: 10.1109/intercon63140.2024.10833498

Resumen

This study addresses the challenge of customer churn in the telecommunications sector by proposing a hybrid model that integrates Machine Learning (ML) and Deep Learning (DL) techniques. Fourteen models were evaluated, including eight individual and six hybrid models, using a public Kaggle dataset across two different scenarios. In the scenario without feature selection, the individual LGBM and XGB models excelled with accuracies of 97.7%, while the hybrid model combining XGB, SVM and RF stood out by achieving a 99.12% accuracy. In the scenario with feature selection, the individual LGBM and XGB models reached an accuracy of 97%, but the hybrid model of XGB, SVM and RF improved to 98.4%. These results surpass several models described in prior literature, demonstrating a significant improvement in churn prediction and providing telecommunications companies with an advanced tool for retaining their customers.

Instituciones: Universidad Peruana de Ciencias Aplicadas