Product Demand Forecasting in E-Commerce with Big Data Analytics: Personalization, Decision Making and Optimization

Abstract

This study explores the role of Big Data in forecasting product demand in the e-commerce sector through the application of machine learning and time series methods. A quantitative descriptive approach is used, involving data collection, preprocessing, analysis, and model evaluation. Forecasting techniques applied include ARIMA for time series prediction and XGBoost for supervised learning to identify key demand factors. Model performance is evaluated using accuracy metrics such as RMSE, MAE, and MAPE. The results indicate that the XGBoost model provides the highest forecasting accuracy at 89%, while the ARIMA model achieves 78%. These findings demonstrate that Big Data significantly supports strategic decision-making in e-commerce by enhancing personalization, optimizing inventory, and enabling data-driven marketing strategies.