Research On LSTM-Based Pricing Strategy for Vegetable Category Wholesale Price Prediction and Multi-Objective Planning

Abstract

Fresh vegetable products constitute an indispensable facet of both daily life and the developmental trajectory of the food industry. Owing to the susceptibility to damage and spoilage, vegetable commodities frequently engender substantial financial losses for food retail establishments. This paper addresses the replenishment and pricing decision problem for vegetable commodities and proposes a solution based on goal programming and LSTM time series forecasting models. Firstly, historical data analysis reveals a linear or logarithmic correlation between vegetable sales volume and cost-plus pricing. Subsequently, an LSTM neural network model is employed to forecast wholesale vegetable prices for the upcoming week. Building upon the forecasted results, a multi-objective programming model is established with the objective of maximizing revenue, utilizing daily replenishment total and pricing as decision variables. The model takes into account the constrained relationship between sales volume and pricing, as well as the lower limit constraint on wholesale prices. Additionally, the paper solves the programming model using the least squares method to determine the optimal daily replenishment quantity and pricing for each vegetable category. Finally, the computational results validate the effectiveness of the proposed method in guiding replenishment and pricing decisions for supermarkets, achieving revenue maximization. This study provides a feasible quantitative decision-making solution for supermarkets facing complex replenishment and pricing issues.