Modeling Customer Behavior in Online Stores Based on the RFM Model and Random Forest and SVM Algorithms

With the increasing volume of purchase history and user activity in online shops, employing machine learning techniques alongside conventional methods like the RFM model has proven to be an effective approach for analyzing customer behavior. One of the key challenges in this area is accurately identifying important customers and the limited use of vast data for marketing decisions. This paper will propose a hybrid methodology that integrates RFM scores with supervised machine learning models—Random Forest and Support Vector Machine (SVM)—to provide a precise method for classifying online store customers. For this purpose, actual data were retrieved from the Kaggle website, and after processing, RFM values were calculated. Subsequently, Random Forest and SVM algorithms were utilized to categorize customers into high-value, medium-value, and low-value segments. The results highlighted that the Random Forest model achieved an accuracy of ۹۲.۶, while the SVM model reached ۹۰.۳۸, indicating strong performance in customer classification. The performance of the models was assessed based on data mining metrics such as accuracy, precision, recall, F۱-score, and AUC. The hybrid model proposed here can effectively support marketing decisions, enhance customer experience personalization, and increase conversion rates and client loyalty in e-commerce environments.

Keywords:

Customer Classification , Supervised Learning Algorithms , Purchase Behavior Prediction , Machine Learning , Classification Performance Evaluation