Performance Evaluation of Supervised Machine Learning Algorithms for Customer Classification in E-Commerce

Customer classification plays a critical role in e-commerce, enabling businesses to personalize services, enhance customer engagement, and improve marketing strategies. With the growing availability of customer data, supervised machine learning (ML) algorithms have emerged as powerful tools for analyzing and predicting customer behavior. This study presents a comprehensive performance evaluation of several popular supervised ML algorithms—namely Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Naive Bayes (NB)—applied to a real-world e-commerce dataset. Key preprocessing steps, including data cleaning, feature selection, and normalization, are conducted to enhance model accuracy. The algorithms are assessed using standard evaluation metrics such as accuracy, precision, recall, F۱-score, and ROC-AUC. Experimental results indicate that ensemble methods, particularly Random Forest, outperform other classifiers in terms of both predictive accuracy and robustness. The findings provide practical insights for e-commerce stakeholders seeking to implement effective customer segmentation strategies. This research contributes to the understanding of algorithmic performance in real-life e-commerce environments and highlights the trade-offs between various classification models. Future work may explore deep learning methods or hybrid approaches for improved classification in more complex customer datasets.

کلیدواژه ها:

Customer Classification ، E-Commerce ، Supervised Machine Learning ، Performance Evaluation ، Predictive Modeling ، Classification Algorithms