Customer Churn Prediction Using Artificial Neural Networks: A Real-Time Web-Based Approach

Abstract

Customer churn prediction is a critical challenge for businesses aiming to retain customers and improve profitability. This paper presents a real-time customer churn prediction system using Artificial Neural Networks (ANN) integrated into a web-based application. The proposed system analyzes customer demographic and behavioral data to identify patterns associated with churn. Data preprocessing techniques such as normalization and encoding are applied to improve model performance. The ANN model is trained using a telecom customer dataset and optimized using the Adam optimizer with binary cross-entropy loss. The system is implemented using a full-stack architecture, where the frontend is developed using React, and the backend is built using Node.js and Express, with the machine learning model integrated via API. Experimental results show that the proposed ANN model achieves high accuracy and outperforms traditional machine learning algorithms in handling non-linear relationships. The developed system provides a scalable and efficient solution for real-time churn prediction and can assist businesses in making proactive retention strategies.

Full Paper

Cite this paper