Research on Vegetable Sales Prediction and Strategy of Supermarkets Based on XGBoost and Random Forest Modeling

This paper mainly focuses on the forecasting and strategy development of vegetable sales in supermarkets. Statistical tools were used to organize and categorize the sales volume and sales unit price of each category, and descriptive statistics and data visualization were completed to observe and analyze the sales distribution pattern of each category and single product of vegetables. First, the association and correlation between individual items and between categories were analyzed by the Spearman correlation coefficient test. Secondly, the optimization model, XGBoost model, and random forest model were constructed to predict the sales volume and price of vegetables in six categories, and the performance of the models was verified by several evaluation indexes. Finally, the objective planning model for maximizing the benefits of the superstore was established, the PSO algorithm was used for optimization, and the daily replenishment quantity and pricing strategy for the coming week were formulated for each category, which provided the superstore with decision-making references with practical value.