Churn Prediction of Customers in a Retail Business using Exploratory Data Analysis

Abstract

These days retail stores and supermarkets are rapidly increasing, and they became very high saturated business. Due to its rapid growth, retail sector is facing very serious problems of customer attrition and churns. So, to overcome this problem, the retail stores and supermarkets need to have an effective churn management strategy. Machine learning, and Data Mining can be used by the management to analyze the churning behavior of customers and help them to retain their customers. To do so, this paper executed explorative data analysis and feature engineering on retail store data set. Five different techniques have been applied namely, Logistic Regression, Random Forest, Decision Tree, K nearest neighbors and XGboost, while Precision, Accuracy, AUC, F1-Score and Recall been used to analyze the performance of classification techniques. This study shows that the proposed model can predict the customer churn with an accuracy of 73% and help management to retain their customers. It is demonstrated in the result that the XGboost is the most efficient classifier for this data set which surpassed all other classifiers in all performance evaluation metrics.