ENHANCING DATA SCIENCE FOR PREDICTIVE ANALYTICS AND PERSONALIZATION IN E-COMMERCE PLATFORMS

Abstract

This study explores the enhancement of e-commerce platforms through data-driven predictive analytics and personalization, leveraging techniques like Matrix Factorization (SVD) and DBSCAN clustering to improve recommendation systems. By analyzing user behavior and transactional data, the proposed methodology achieves a 78.5% precision, 72.1% recall, and 75.2% F1-Score, demonstrating robust relevance in recommendations. Deployment results show a 124% increase in conversion rates (from 2.1% to 4.7%) with SVD, further improving to 5.3% when combined with DBSCAN. The framework addresses challenges like data sparsity and the processing, offering scalable solutions for personalized marketing. These outcomes highlight the potential of hybrid models to drive customer engagement and revenue growth in competitive e-commerce environments.