A Product Replenishment Decision Algorithm Based on ARIMA-RF and Spearman Analysis

Reliable market demand analysis and reasonable sales mix are crucial to the replenishment of fresh food superstores, and this paper analyzes the sales pattern of different categories and single products of vegetables and forecasts the sales of vegetables by establishing Spearman correlation analysis, ARIMA-RF and other models. Firstly, Spearman correlation analysis is used to search for the correlation between vegetable categories and single products respectively, and it is found that there is no strong correlation between different categories and different single products. Then heat map, matrix scatter plot and 2D scatter plot were used to visualize the integrated category weekly data, and combined with Lyapunov index to make a preliminary judgment, and it was found that all categories might be nonlinear and chaotic in nature. Finally, ARIMA-RF model is developed to predict the daily replenishment for the coming week and its prediction interval is calculated. We found that the ARIMA-RF mixed model has a good effect on predicting the price and sales of different types of vegetables, and can provide theoretical support for the supermarket to carry out the replenishment strategy of different types of vegetables.