Leveraging Ensemble Learning with Deep Learning for Accurate Customer Churn Prediction in Subscription-Based Mode

Document Type

Conference Proceeding

Publication Date

1-24-2025

School

Computing Sciences and Computer Engineering

Abstract

An ensemble learning approach to address customer churn prediction for businesses which utilize subscription models. Ensemble methods, including bagging, boosting, and stacking, are combined with multiple base learners to improve predictive accuracy, leveraging deep learning architectures like convolutional and recurrent neural networks to capture temporal and spatial dependencies in customer behavior. The research utilizes a comprehensive dataset comprising historical customer interactions, demographic details, service usage, and churn outcomes. Preprocessing ensures the dataset’s readiness for machine learning. Ensemble models aggregate predictions from multiple classifiers, mitigating overfitting risks and achieving superior performance in precision, recall, and F1 score compared to individual models. Thus, the proposed methodology helps to enhance customer lifetime value, promote long-term business development and increase customer satisfaction with 98% accuracy. This approach offers more reliable churn predictions and actionable insights, enabling businesses to design targeted retention strategies and optimize services. By maximizing customer lifetime value, the proposed methodology contributes to sustained business growth and improved customer satisfaction.

Publication Title

International Conference on Intelligent Systems and Computational Networks, ICISCN 2025

Share

COinS