The Research Department of Economics and Management of Tadbir Nikan Business, Marketing, and Finance Open 3092-6238 In Press 2027 01 01 Comparison of Classification Algorithm Performance in Predicting Customer Churn in Retail Stores Based on the RFM Model Comparison of Classification Algorithm Performance in Predicting Customer Churn in Retail Stores Based on the RFM Model 1 19 EN Journal Article 2025 12 14 <p>Predicting customer churn is considered one of the fundamental issues in the retail industry because a decline in customer retention rates directly affects acquisition costs and profitability. Accordingly, the present study was conducted with the aim of developing a comprehensive analytical framework for customer churn prediction based on the behavioral RFM model (Recency, Frequency, Monetary) by utilizing various machine learning algorithms. In this study, actual customer transaction data from a retail store were used, and the algorithms of Logistic Regression, Decision Tree, Support Vector Machine (SVM), Naïve Bayes, AdaBoost, and the hybrid Stacking model were evaluated and compared. Following normalization, data cleaning, and feature engineering processes, the models were assessed using the quantitative metrics of Precision, Accuracy, Recall, and F1-Score. The evaluation results demonstrated that the Stacking model outperformed the other algorithms and, with an overall accuracy of approximately 87%, provided a highly accurate framework for predicting churn probability. Variable importance analysis indicated that the Recency indicator plays the most significant role in predicting churn tendency, while the combination of Frequency and Monetary indicators improves model stability. In addition, K-Means clustering and association rule techniques (FP-Growth) were employed for behavioral analysis and customer segmentation, and the results emphasized the effectiveness of combining purchasing pattern analysis with RFM segmentation to improve churn prediction model accuracy. The primary innovation of this study lies in the integration of three analytical approaches—association rules, Stacking classification, and RFM clustering—within a unified empirical framework that establishes a systematic relationship among customer segmentation, purchasing behavior, and churn probability. This framework can serve as a practical basis for designing intelligent customer retention campaigns in the retail industry.</p> https://bmfopen.com/index.php/bmfopen/article/download/434/482