Unveiling Customer Behavior: A Comprehensive Study on Predictive Models for Telcom Churn
Authors:
- 5 March 2025
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Abstract
Predicting churn in telecoms with a range of Random Forest algorithms is a necessary and systematic way of foreseeing customers’ intention to leave the service. The research presented here focuses on telecom churn prediction by using the Random Forest algorithm, whose main objective is the early prediction of customer rejection probability. When telecom churn is indicative of how often customers abandon their relationships with companies or cut ties from service subscriptions, it has become an important subject for telecom firms to accurately predict it and best avoid it. To achieve this, a publicly available telecom customer dataset from [mention dataset source, e.g., Kaggle or UCI Machine Learning Repository] was utilized. The dataset comprises 20 independent variables and 1 target variable, with a total of 7043 rows. The data was split into a 70:30 ratio, where 70% of the data was used for training and 30% for testing. A Random Forest algorithm was applied and assessed for its capability in this predictive task. This analysis concentrates on assessing predictive accuracy, e.g., precision, recall, and F-1 scores, to determine whether the algorithm is effective in detecting potential churners early. Moreover, our study of interpretability also involves the identification of feature importance in the Random Forest framework to unveil some of the factors contributing to customer churn. The results of this paper highlight the main factors influencing churn prediction and serve as a strong foundation for telecom companies to address the problem of consumer churn. Embracing the predictive power of Random Forest could change the pace at which telecom companies lose customers, improve customer satisfaction, and ensure responsible resource utilization, ultimately fostering a sustainable business environment in a competitive market.Keywords
This publication has 17 references indexed in Scilit:
- Using machine learning methods to predict future churners: an analysis of repeat hotel customersInternational Journal of Contemporary Hospitality Management, 2024
- Churn prediction analysis of telecom customers using svm, random forest and logistic regression models using orange data mining toolsE3S Web of Conferences, 2024
- Empirical analysis of tree-based classification models for customer churn predictionScientific African, 2023
- Customer churn prediction in telecom sector using machine learning techniquesResults in Control and Optimization, 2023
- A classification application for using learning methods in bank costumer's portfolio churnJournal of Forecasting, 2023
- An improved random forest based on the classification accuracy and correlation measurement of decision treesExpert Systems with Applications, 2023
- Identifying SH-IoT devices from network traffic characteristics using random forest classifierWireless Networks, 2023
- Supervised Classification Algorithms in Machine Learning: A Survey and Review2019
- An up-to-date comparison of state-of-the-art classification algorithmsExpert Systems with Applications, 2017
- An improved random forest classifier for multi-class classificationInformation Processing in Agriculture, 2016