Recent Advances in E-Commerce System for Sale Prediction Using Triple Pseudo-Siamese Network with Giant Trevally Optimizer: A Systematic Review

Abstract

The rapid expansion of digital commerce platforms has led to the generation of massive volumes of transactional and behavioral data, creating new opportunities for advanced analytics in retail decision-making. Sales prediction has become a critical application, enabling businesses to optimize inventory management, demand forecasting, pricing strategies, and supply chain operations. Traditional forecasting methods such as regression models, ARIMA, and exponential smoothing have been widely used; however, they often fail to capture the complex nonlinear patterns and dynamic nature of modern e-commerce data. Recent advancements in machine learning and deep learning, including random forests, gradient boosting, convolutional neural networks (CNNs), long short-term memory (LSTM) networks, and transformer models, have significantly improved prediction accuracy by learning intricate relationships within large datasets. Among emerging approaches, Siamese neural networks have gained attention for their ability to model similarity across heterogeneous inputs, enhancing forecasting by analyzing relationships among product attributes, customer behavior, and historical sales. Additionally, nature-inspired optimization techniques such as the Giant Trevally Optimizer (GTO) improve model performance and convergence. This review examines recent studies (2020–2023), highlighting hybrid models that integrate deep learning and optimization for robust e-commerce sales prediction.