Predicting Customer Churn Based on Deep Learning, Neural Networks and Logistic Regression

Abstract

Today's competitive world has made keeping old customers one of the most important goals of economic enterprises. Customer attrition management in order to minimize losses from customer attrition and maximize profits from retaining valuable customers, as a powerful tool, analyze customer behavior using available data and identify and target customers prone to attrition. This category of customers plans and implements appropriate and effective strategies to maintain them. Therefore, the main goal of this study is to predict customer churn in Koroosh chain stores in Tehran by using the deep learning method and comparing it with the results of neural networks and logistic regression. The results of the study showed that the deep learning model has succeeded in better classification and prediction than the other introduced models, according to the key criteria of model evaluation.