Automatic Pricing and Replenishment Strategy for Vegetable Products Based on Time-Series Analysis and BP Neural Fitting

Abstract

Listen

Translate article

To reduce the waste of vegetables and guarantee revenues for supermarkets, this paper explores the construction of a comprehensive optimization model to help the supermarket select the specific suitable replenishment scheme and pricing strategy. This sets the bedrock for it to both gain the maximum revenue and timely supply various categories of vegetables that the market demands. The expectation for further relevant research to better complete the model is also mentioned. First, the attached data were preprocessed, and no missing values and outliers were detected. Consequently, the data of multiple forms was merged, and data related to categories of vegetables was counted. In response to Question 1, descriptive statistics were conducted on different individual items and categories of vegetables. Sales volumes of categories of vegetables were visualized through sales volume statistical graphs, profit line graphs, and quarterly sales change graphs. Following, the sales volumes were verified to have satisfied the conditions for calculating the Pearson Correlation Coefficient, and heat maps were drawn for correlation analysis. For individual items, hierarchical clustering was carried out with indicators, such as sales volume, unit price, number of purchases, and wastage rate. The basis of categorization of each category of vegetable was also explored. For Question 2, average pricing was used to replace cost-plus pricing first. Then, BP Neural Net Fitting was leveraged to analyze the relation between total sales volume and average pricing of different vegetable categories. The average wholesale price of the next seven days of each vegetable category was predicted with ARIMA Model, in order to gain the profit of different categories. Finally, a nonlinear objective planning model to achieve maximum benefit for the supermarket was constructed. Corresponding constraints were given to propose a reasonable total replenishment and pricing strategy for each vegetable category. In solving Question 3, constraints were added based on the nonlinear objective planning model in Question 2, and the prediction model was optimized. In the case of meeting market demand, the replenishment volume and pricing strategy for individual items on July 1, 2023, were proposed based on a combination of factors, as a way to maximize the benefits for the supermarket.