Precise E-Commerce Information Push Model Based on User Feature and Improved Stacking

Precise information push plays a critical role in the growth of the e-commerce sector. To overcome limitations of traditional recommendation algorithms, such as low prediction accuracy and suboptimal push performance, this study proposes an e-commerce information push model based on user feature integration and an improved Stacking ensemble framework. The proposed model employs Random Forest, Logistic Regression, and Extreme Gradient Boosting as base learners and Multiple Linear Regression as the meta-learner. By incorporating user feature information, the framework generates personalized product recommendations with enhanced predictive performance. Experimental results demonstrate that the improved Stacking model achieves a root mean square error (RMSE) of 7.21 on the test set, outperforming comparison algorithms in both stability and accuracy. When evaluating recommendation lists of ten items, the model achieves a normalized discounted cumulative gain (nDCG) of 0.17, significantly higher than baseline approaches, indicating superior push performance. In summary, the proposed e-commerce information push model outperforms existing methods in prediction accuracy, stability, and recommendation quality, providing a robust framework to support more precise and effective information push strategies for e-commerce platforms.