Time Series Based Decision Making Problem of Vegetable Replenishment and Pricing in Superstores

With the progress of living standards, people are more and more thirsty for green and healthy life, for the purchase of vegetables also has a higher pursuit of hope to buy fresh and cheap vegetables, and due to the short freshness period of vegetables, seasonal characteristics, super sales space is very limited. In this paper, through the quantification of the previous sales data, we explored the relationship between vegetable categories and analyzed the vegetable pricing strategy. First of all, the sales volume of vegetable categories is analyzed from the perspective of year, season and month, and the correlation is verified by Spearman's test, and the sales volume of individual products is divided according to the month, and then clustering analysis is carried out by using K-Means cluster analysis based on PSO particle swarm algorithm. Then, the final replenishment quantity is obtained by time series forecasting using the combined time series model based on LSTM-XGBoost, the average wholesale price is obtained by using the ARIMA time series forecasting model, a nonlinear planning model is built out, and finally the single-item sales strategy is derived by a heuristic algorithm. Finally, taking whether to stock the vegetable on the shelves on a single day as the decision variable, this paper maximizes revenue as the objective function, defines the sales space and the loss rate threshold as the constraints, establishes a 0–1 planning model, and solves to obtain the single-day sales strategy by using the heuristic algorithm. The vegetable replenishment and pricing model established in this paper can be closely linked with the actual situation, can be combined with the actual situation to solve the problem, the model has good generality and popularization.