Customer churn prediction using composite deep learning technique.

Area of Science:

• Computer Science
• Artificial Intelligence
• Business Analytics

Background:

• Customer churn poses significant financial challenges for businesses, impacting customer retention efforts.
• Existing machine learning and deep learning models often struggle with accurate customer churn prediction.
• Previous methods overlooked sequential information in deep neural network feature extraction.

Purpose of the Study:

• To develop an effective hybrid deep learning model for accurate customer churn prediction.
• To enhance the accuracy and reliability of customer churn estimation processes.
• To address limitations in current ML/DL approaches for churn detection.

Main Methods:

• A novel hybrid deep learning model, BiLSTM-CNN, was developed.
• The model integrates Bidirectional Long Short-Term Memory (BiLSTM) and Convolutional Neural Network (CNN) components.
• The proposed model was trained, tested, and validated on a benchmark dataset.

Main Results:

• The BiLSTM-CNN model achieved a remarkable prediction accuracy of 81% on the benchmark dataset.
• Experimental results demonstrate the model's effectiveness in estimating customer churn.
• The hybrid approach shows improved performance over traditional methods.

Conclusions:

• The BiLSTM-CNN model offers a promising and effective solution for predicting customer churn.
• This advanced deep learning technique can significantly improve customer retention strategies.
• The study highlights the importance of sequence information in churn prediction models.