E-Ticarette Kullanıcılarının Davranışsal Yaklaşımlarının Makine Öğrenmesi Yöntemleri ile Sınıflandırılması ve Müşteri Segmentasyonu
Classification of Users' Behavioral Approaches in E-Commerce and Customer Segmentation Using Machine Learning Methods

Summary/Abstract: The rapid growth of e-commerce and increasing consumer expectations have intensified the need for efficient customer segmentation methods. Traditional segmentation techniques, which often rely on manual and rule-based approaches, fail to address the complexity and scalability required in modern e-commerce. Machine learning algorithms provide a powerful alternative, offering data-driven insights that enable personalized marketing strategies and targeted customer engagement. This study aims to analyze and compare the effectiveness of K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Random Forest algorithms in the classification of customer segments in e-commerce. Existing literature suggests various approaches for customer segmentation, but a direct comparison of these machine learning models in an e-commerce context remains limited. This research seeks to fill this gap by evaluating the classification performance of these models and determining their suitability for segmenting online shoppers. The dataset utilized in this study was obtained from Kaggle and contains consumer behavior-related features, including shopping frequency, promotional engagement, spending habits, and security concerns. To determine the optimal number of clusters, Elbow and Silhouette methods were employed. Once the clustering structure was established, the K-Means algorithm was applied to segment customers into distinct groups based on their behavioral patterns. These clusters were then classified using KNN, SVM, and Random Forest, and their performance was assessed using standard classification metrics such as Accuracy, Precision, Recall, and F1- Score. The results revealed that SVM achieved the highest accuracy (95%), making it the most reliable model for customer segmentation. Random Forest closely followed with an accuracy of 93%, demonstrating strong performance while offering higher scalability and lower computational costs. KNN, on the other hand, showed the lowest accuracy (78%), indicating that it may not be the most suitable choice for highdimensional or large datasets. In addition, SVM and Random Forest consistently outperformed KNN in terms of Precision, Recall, and F1-Score, confirming their superior classification capabilities. In conclusion, this study highlights the importance of selecting the right machine learning model for customer segmentation in e-commerce. SVM is recommended for applications requiring the highest classification accuracy, while Random Forest is better suited for large datasets with high scalability demands. KNN, despite its simplicity, is more appropriate for small-scale segmentation tasks. The findings provide valuable insights for ecommerce businesses seeking to optimize their segmentation strategies through machine learning, ultimately enhancing customer targeting, engagement, and retention.

• Details
• Contents