This project focuses on creating a machine learning-based system for customer segmentation and product recommendations using data from e-commerce platforms. The goal is to analyze customer behaviors and purchasing patterns to enhance the personalization of product suggestions and improve overall customer experience. Existing systems typically rely on basic segmentation methods, such as demographics or past purchases, which can overlook subtle customer preferences and lack real-time data integration. The proposed system utilizes advanced ensemble learning techniques, particularly stacking models, to improve prediction accuracy. The approach includes data preprocessing, feature engineering, and the application of models like Random Forest, XGBoost, and LightGBM. A stacking ensemble model combines the strengths of these algorithms, while hypergraph models capture complex relationships between customers and products. Anomaly detection techniques help identify outliers in customer behavior, further refining the segmentation process. Explainable AI tools such as SHAP and LIME enhance model transparency. Additionally, the project incorporates time series forecasting to predict customer trends and employs Natural Language Processing (NLP) for sentiment analysis of reviews. Federated learning techniques are used to ensure data privacy during collaborative model training across decentralized datasets. The results indicate significant improvements in accuracy and segmentation effectiveness, with the ensemble model, especially the stacking approach, surpassing individual models. This system offers a more advanced and personalized method for customer segmentation compared to traditional techniques, with promising applications for real-time recommendation systems in e-commerce.