Research on Pricing and Replenishment Strategies of Supermarkets Based on PSO-LSTM- Transformer Models

Abstract

The supermarket's role in ensuring market supply and enhancing livelihoods prompted the implementation of a new PSO-LSTM-Transformer hybrid model in this study. This amalgamation combines Particle Swarm Optimization (PSO), Long Short-Term Memory (LSTM), and Transformer to forecast commodity prices and inventory needs. By combining these methods, the research aimed to satisfy market requirements while maximizing grocery store profits and improving market operations sustainability.Particle Swarm Optimization (PSO) enhances prediction accuracy by effectively exploring the vast solution space for optimal solutions. Long Short-Term Memory (LSTM), which is known for capturing long-term data dependencies, enhances the comprehension and forecasting of market trends. Furthermore, the Transformer model enhances the forecasting process by capturing intricate patterns and relationships in market data through its attention mechanism.The research concludes that the PSO-LSTM-Transformer model effectively predicts commodity pricing and replenishment needs, thereby aiding supermarkets in making informed decisions to balance market demands and optimize revenue. The findings of the study contribute to the promotion of forecasting methods within the supermarket sector, facilitating efficient management of the market and sustainable economic growth.