Traditional e-commerce marketing strategies are struggling to keep pace with the need for personalized customer experiences. Customer acquisition costs often outweigh the benefits of retaining existing customers, highlighting the importance of understanding long-term customer value. However, bombarding customers with irrelevant information can backfire, leading to lost sales. Many current marketing techniques lack a focus on Customer Lifetime Value (CLV), use overly simplified segmentation methods, and do not account for how customer behavior evolves. This project seeks to overcome these challenges by developing a comprehensive customer segmentation model for e-commerce: K-means clustering, DBSCAN, and GMM. This project combines data analysis techniques with machine learning algorithms to predict CLV using classification models: Decision Tree, Random Forest, Gradient Boosting, and SVM. Customer segments will be created based on their predicted value and behavioral patterns. The project will then measure the effectiveness of the segmentation and prediction approach, evaluating each machine learning model to know the performance. The result shows that K-Means balanced simplicity and efficacy, while DBSCAN was particularly useful for detecting and removing outliers. Gradient Boosting emerged as the best-performing model, delivering high accuracy and stability across different segmentation methods. Decision Tree and Random Forest also performed well, particularly in GMM-based segmentation, while SVM showed weaker performance with lower accuracy and higher variance. The result of this study could offer e-commerce businesses valuable ways to target marketing, improve customer retention, and boost long-term profitability.