Comparative Performance Analysis of Logistic Regression and Random Forest for Telecommunications Customer Churn Prediction Using Recursive Feature Elimination
DOI:
https://doi.org/10.30998/faktor.exac.v19i1.828Keywords:
churn, feature selection, logistic regression, prediksi, random forestAbstract
Customer retention is a critical concern in the telecommunications sector due to high revenue risks and subscriber acquisition costs. While machine learning is widely used for churn prediction, cross-validated studies evaluating the impact of Recursive Feature Elimination (RFE) on Logistic Regression and Random Forest remain limited. This research compares these two algorithms, selected for interpretability and non-linear modeling capabilities respectively, before and after RFE feature selection using the Telco Customer Churn dataset (7,043 records). Performance was assessed via 5-fold cross-validation using accuracy, recall, precision, and F1-score. RFE isolated ten key predictors (SeniorCitizen, Dependents, Tenure, PhoneService, MultipleLines, OnlineSecurity, OnlineBackup, TechSupport, Contract, and PaperlessBilling), significantly reducing the feature space. Experimental results show that RFE increased Logistic Regression accuracy from 73.46% to 79.61% (F1-score: 42.35% to 58.60%). Conversely, Random Forest exhibited a modest accuracy gain from 79.85% to 81.00% (precision: 71.00%, F1-score: 63.85%). Ultimately, RFE yields more substantial improvements for Logistic Regression, though Random Forest delivers the highest overall predictive performance.
Downloads
References
[1] A. Amin, S. Anwar, A. Adnan, M. Nawaz, N. Howard, and J. Qadir, “Customer churn management in the telecommunications sector: A predictive analytics approach,” J. Bus. Analytics, vol. 7, no. 2, pp. 112–129, 2024.
[2] E. Shaaban, Y. Helmy, A. Khedr, and M. Nasr, “A proposed churn prediction model,” Int. J. Comput. Appl., vol. 975, no. 8887, 2020.
[3] Selular.id. (2024, July). “Tingkat churn operator seluler Indonesia capai 26% pada Juli 2024,” Selu-lar.id, Jul.2024[online]. Available: https://selular.id
[4] A. Idris, M. Rizwan, and A. Khan, “Churn prediction in telecom sector using various machine learning algorithms,” Comput. Electr. Eng., vol. 76, pp. 272–285, 2019.
[5] A. Mandurkar and S. Khanke, “Churn Prediction Using Various Machine Learning Algorithms and Feature Selection,” 2022 International Conference on Emerging Trends in Smart Technologies (ICETST), 2022.
[6] R. Patil, “Customer Churn Prediction using Machine Learning in E-commerce,” International Journal of Innovative Research in Computer Science & Technology, 2024.
[7] P. Raj and S. Gupta, “The Efficacy of Customer Churn Prediction System Using Different Machine Learning Models,” International Journal of Scientific & Technology Research, 2024.
[8] R. Dodda and S. Raghavendra, “A Comparative Study of Machine Learning Algorithms for Custom-er Churn Prediction,” International Conference on Artificial Intelligence and Data Engineering (AIDE), 2024.
[9] S. Pulkundwar and A. Rudani, “A Comparison of Machine Learning Algorithms for Customer Churn Prediction,” International Journal of Computer Science and Mobile Computing, vol. 12, no. 4, pp. 45–54, 2023.
[10] L. Breiman, “Random forests,” Mach. Learn., vol. 45, no. 1, pp. 5–32, 2001.
[11] I. Guyon and A. Elisseeff, “An Introduction to Variable and Feature Selection,” Journal of Machine Learning Research, vol. 3, pp. 1157–1182, 2003.
[12] E. Mufida, D. Andriansyah, and H. Hertyana, “Customer churn prediction pada sektor perbankan dengan model Logistic Regression dan Random Forest,” Comput. Sci. (CO-SCIENCE), vol. 5, no. 1, 2025.
[13] S. Antoh, R. Herteno, I. Budiman, D. Kartini, and M. M. Itqan, “Prediksi churn pelanggan telekomu-nikasi dengan optimalisasi seleksi fitur dan tuning hyperparameter pada algoritma klasifikasi C4.5,” J. Syst. Inf. Bus., vol. 15, no. 1, 2025.
[14] Holilurrahmna & Mohamad Imron. “Implementasi Model Prediksi Churn Pelanggan menggunakan Algoritma Random Forest Pada Website Industri Telekomunikasi”, Jurnal Teknologi Informasi (AJTI)., vol.1, no.1, 2025.
[15] D. McMurray, “Telco Customer Churn Dataset,” Kaggle Datasets, 2018.
[16] M. Hofmann and R. Klinkenberg, RapidMiner: Data Mining Use Cases and Business Analytics Appli-cations. Boca Raton, FL: CRC Press, 2013.
[17] A. Kotu and D. Deshpande, Predictive Analytics and Data Mining: Concepts and Practice with RapidMiner, 2nd ed. Waltham, MA: Morgan Kaufmann, 2019.
[18] F. Admojo and S. Jabir, "Analisis Performa Metode Naive Bayes Clasifier pada Eectronic Nose Dalam Identifikasi Formalin Pada Tahu," Indones.J. Data SCI, vol. 4, pp. 1-16, 2023.
[19] R. R. Sudriyanto and M. A. R. Hariri, "Implementasi Algoritma Decision Tree (C4.5) dengan Optimize Wheights (PSO) Untuk Memprediksi Kelulusan Mahasiswa Tepat Waktu," Jurnal Informatika Universitas Pamulang, vol. 6, pp. 252-257, 2021.
[20] Hosmer, D. W., Lemeshow, S., & Sturdivant, R. X. (2013). Applied logistic regression. John Wiley & Sons.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Anggreita Tiara Putri, Aa Kurniawan, Nurfiqih Nurfiqih (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Under the CC BY-NC 4.0 License, users are permitted to:
-
Share: Copy, distribute, and transmit the material across any medium or format.
-
Adapt: Remix, transform, and build upon the work for non-commercial purposes.
These permissions are granted provided the following conditions are met:
-
Attribution (BY): Proper credit must be given to the original author(s) and Faktor Exacta, including a direct link to the license and a clear indication of any modifications made. This acknowledgment must not imply endorsement by the journal or the authors. Reusers are required to include the article’s Digital Object Identifier (DOI) for permanent citation reference.
-
NonCommercial (NC): The material may not be exploited for commercial purposes or monetary gain. Any commercial utilization, republication, or integration into commercial goods requires explicit, prior written permission from the copyright holder(s).
-
No Additional Restrictions: No legal provisions or technological protection measures (such as DRM) may be applied to restrict others from exercising the freedoms granted by this license.
Official Legal Code: For the full and legally binding terms, please review the official Creative Commons Attribution-NonCommercial 4.0 International Public License.
