Archimedes Optimization Algorithm-Based Feature Selection with Hybrid Deep-Learning-Based Churn Prediction in Telecom...

Area of Science:

• Data Science and Machine Learning
• Telecommunications Analytics

Background:

• Customer churn prediction (CCP) is vital for businesses to retain subscribers and ensure profitability.
• Existing methods often struggle with high-dimensional data and optimal feature selection in telecom.
• Deep learning (DL) offers potential for robust predictive models but requires careful optimization.

Purpose of the Study:

• To develop an efficient hybrid deep-learning model for customer churn prediction in the telecom industry.
• To address high-dimensionality issues through advanced feature selection techniques.
• To optimize model hyperparameters for enhanced classification performance.

Main Methods:

• Proposed the Archimedes Optimization Algorithm-based Feature Selection with a Hybrid Deep-Learning-based Churn Prediction (AOAFS-HDLCP) technique.
• Utilized the Archimedes Optimization Algorithm (AOAFS) for optimal feature selection.
• Employed a Convolutional Neural Network with Autoencoder (CNN-AE) for the core prediction task.
• Applied Thermal Equilibrium Optimization (TEO) for hyperparameter tuning of the CNN-AE model.

Main Results:

• The AOAFS-HDLCP technique demonstrated superior performance compared to other methods.
• Achieved a maximum classification accuracy of 94.65%.
• Effectively mitigated high-dimensionality problems through optimized feature selection.

Conclusions:

• The proposed AOAFS-HDLCP technique offers a robust and efficient solution for telecom customer churn prediction.
• The hybrid DL approach combined with advanced optimization algorithms significantly improves predictive accuracy.
• This methodology enhances customer retention strategies and contributes to business profitability.