# Classification of customer retention using hybrid SVC-SDNN to enhance customer relationship management

## TL;DR

This paper introduces a hybrid AI model to predict customer churn, improving customer relationship management in banking and corporate sectors.

## Contribution

A novel hybrid SVC-SDNN model is proposed for enhanced customer churn prediction and retention analysis.

## Key findings

- The hybrid model achieves an AUC-ROC of 0.881696 and accuracy of 0.95.
- The model outperforms other hybrid algorithms in predictive metrics like precision and recall.
- The approach identifies key factors influencing customer retention effectively.

## Abstract

Many banking and corporate sector organization problems are resolved by clever, creative solutions based on artificial intelligence (AI). Any financial institution has to use AI-enabled churn detection solutions to improve customer relationship management (CRM). In order to effectively predict churn in a publicly accessible datasets, we suggest a novel hybrid deep method. It functions efficiently on any private, hidden banking dataset in the specified format. Other hybrid algorithms’ performances are contrasted with this one. The predictive analytics of SDNN and SVM coupled is excellent using accuracy, precision, recall, and F1-score matrices, according to our thorough search for the best intelligent solution. It is essential to correctly identify the important variables or churn-causing elements. SVC-SDNN’s strength is in its ability to anticipate which customers are most likely to leave and identify the critical elements affecting customer retention. The suggested approach has an AUC-ROC of 0.881696 and an accuracy of 0.95.