Frontiers in Business, Economics and Management ISSN: 2766-824X | Vol. 11, No. 2, 2023 352 Random Forest‐Based Restocking and Pricing Prediction for Vegetable Items Kaile Wang1, †, Keming Su2, †, Hao Li2, † 1 School of Information Engineering, Inner Mongolia University of Science & Technology, Baotou, China 2 School of Civil Engineering, Inner Mongolia University of Science & Technology, Baotou, China † These authors also contributed equally to this work Abstract: The intolerance to storage of vegetable commodities in supermarkets makes automatic pricing and replenishment decisions for vegetable commodities particularly important. This paper takes the measured data of a superstore as an example to formulate a set of effective pricing and replenishment decisions for vegetable commodities, which is a comprehensive consideration to ensure the balance of supply and demand, and to reduce the losses of the superstore and the loss rate of commodities. First of all, the sales of each category and single product in different time periods were counted, and the Pearson correlation coefficient was calculated to obtain the distribution pattern of the sales volume of each category and single product of vegetables. Then, the relationship between the total sales volume of and the cost-plus pricing of each vegetable category is analyzed, and a random forest model is established to predict the total replenishment volume and pricing strategy in the coming week. Finally, the replenishment quantity and pricing strategy of individual items are given to maximize the revenue of the superstore under the premise of trying to meet the market demand for each category of vegetable goods. The model established in the paper, which basically solves the given problem, has strong practicality and high computational efficiency. Keywords: Pricing Strategies, Pearson Correlation Coefficient, Regression, Random Forest. 1. Introduction In recent years, with the development of our country, the market demand for fresh products is also expanding. However, the problem has also emerged, in China's fresh food industry chain, it will be due to the mass demand, supply capacity, price discomfort and other reasons to cause problems such as stagnant sales of fresh food, fresh food in short supply, and cause loss of revenue and waste of commodities and other problems [1-3]. The reason for this phenomenon is that the sales strategy is not suitable, the supply and demand is not enough to understand, and the pricing mechanism is unreasonable. At present, most of the domestic supermarket fresh sales, the whole process of fresh sales for the price of the same, the approach of the low price sold at the end of the period, resulting in a large number of fresh rot and stagnation, not only let the supermarket and fruit and vegetable dealers to suffer serious economic losses, but also make the fresh to cause unnecessary waste. With the improvement of people's living standards and the rapid dissemination of information, consumers show diversity in their purchasing decisions, focusing not only on the value and cost-effectiveness of goods, but also on the quality and safety of goods. The intolerance to storage of vegetable products in supermarkets makes automatic pricing and replenishment decisions for vegetable products particularly important, which involve the influence of many factors such as market demand, supply, and cost. In order to ensure effective pricing and replenishment decisions for vegetables, as well as to improve the shopping experience of customers, we need to take into account to ensure the balance between supply and demand, so as to reduce the loss of superstores and the loss rate of goods [4-6]. In order to solve the above problems, this paper takes the measured data of a superstore as an example, firstly, to find out the distribution law of the sales volume of each vegetable category and single product and their interrelationship. Then, analyze the relationship between the total sales volume of each vegetable category and the cost-plus pricing, and give the total daily replenishment and pricing strategy of each vegetable category in the coming week, so that the superstore can maximize the revenue. Finally, due to the limited sales space of vegetable items in the superstore, a replenishment plan for new individual items is formulated to maximize the revenue of the superstore under the premise of trying to satisfy the market's demand for vegetable items in each category. 2. Sales Volume Distribution Pattern and Correlation Analysis The measured data of a hypermarket gives the categories to which all vegetables in the hypermarket belong and the sales flow of each individual item. Analyzing and processing the measured data of a superstore, we get the sales of each category and single product in different quarters . According to the sales situation in different time periods, the distribution pattern of the sales volume of each category and single product of vegetables can be inferred. The change in sales (in kilograms) of individual dishes from the third quarter of 2020 to the second quarter of 2023 is shown in Table.1. 353 Table 1. Statistics on total sales by category in each quarter kind season philodendron cauliflower Aquatic rhizomes eggplant capsicum edible fungi 2020 Q3 6589.717 2535.512 7281.936 2984.788 7012.806 6149.678 2020 Q4 2314.155 2919.946 9250.989 1354.473 5174.008 11643.83 2021 Q1 3642.985 2059.854 9853.369 1931.269 8138.264 11004.17 2021 Q2 6124.785 2172.837 6411.767 2010.647 6187.02 4621.28 2021 Q3 6334.069 2155.516 8315.872 2513.558 4888.864 5101.016 2021 Q4 2260.045 2035.208 9391.983 913.862 3102.582 7343.637 2022 Q1 2293.363 2746.059 9149.283 1707.858 5463.779 5562.767 2022 Q2 4578.351 1890.901 6234.736 2576.898 1556.909 3352.926 2022 Q3 2160.394 3527.74 9149.877 1299.507 7760.977 6499.568 2022 Q4 474.35 2129.833 10489.672 623.969 7931.845 11748.3 2023 Q1 740.752 2026.572 9181.67 1438.505 9550.94 10947.53 2023 Q2 2558.88 1338.122 6724.201 2028.759 6222.956 6850.968 In order to more graphically represent the volume of sales in each category for each season, the seasonal sales volume is plotted as shown in Figure 1. Figure 1. Quarterly sales of six major types of vegetables Assuming that the sales volume of the two highest-selling individual products in each category of vegetables reflects the distribution pattern and interrelationship of the sales volume of individual products in that category, and because it is more complicated to write out the interrelationship and distribution pattern of the sales volume of all the vegetable categories in the text, we use eggplant as a representative. The two highest selling items in the eggplant category are: purple eggplant (2) and green eggplant (1), and the line graph of sales of these two eggplants by quarter is shown in Figure 2. Figure 2. Line graph of sales of purple eggplant (2) and green eggplant (1) by quarter 0 2000 4000 6000 8000 10000 12000 14000 2020 Q3 2020 Q4 2021 Q1 2021 Q2 2021 Q3 2021 Q4 2022 Q1 2022 Q2 2022 Q3 2022 Q4 2023 Q1 2023 Q2 philodendron cauliflower Aquatic rhizomes eggplant capsicum edible fungi 0 200 400 600 800 1000 1200 1400 1600 1800 2020 Q3 2020 Q4 2021 Q1 2021 Q2 2021 Q3 2021 Q4 2022 Q1 2022 Q2 2022 Q3 2022 Q4 2023 Q1 2023 Q2 Purple eggplant (2) Green eggplant (1) 354 There are two types of eggplant in each quarter after the sales, according to the Pearson correlation coefficient calculation process, the specific Pearson correlation coefficient calculation formula is as follows: 𝑅 ∑ ∑ (1) The correlation coefficient was calculated using MATLAB: 0.5009. The associated strength rating scale is shown in Table.2. Since R=0.5009 0.6 , it indicates that the positive correlation between the top two selling items in the eggplant category is moderately correlated. Table 2. Table of relevant strength classes Numerical range degree of relevance 0.8-1.0 Highly relevant 0.6-0.8 strong correlation 0.4-0.6 Moderately relevant 0.2-0.4 weak correlation 0.0-0.2 Very weak correlation or no correlation 3. Individual Product Replenishment Program Development The sales volume and selling price of each day of the vegetable category were obtained, and then the data of the first day of the latter 18 months were selected to establish a regression model, so as to derive the relationship between the total sales volume and the average unit price of each vegetable category. It is known that the total daily replenishment is positively correlated with the total sales volume, so the pricing strategy is determined by the positive correlation between the ratio of wholesale unit price and sales unit price. Since the variation of daily sales volume and wholesale price is small, we selected the wholesale price and sales volume in the recent week and predicted the sales volume in the coming week by using the random forest model. Data analysis was carried out to obtain the values of sales and average selling price of cauliflower category as shown in Table 3. Table 3. Cauliflower sales and average unit price sales volume Average unit price 62 8 13.968 10 22.944 10 25.701 8.4 23.295 10 22.234 8 28.897 8.9 82.828 8.4 47.503 12.7 76.779 10.1 39.603 7.8 24.722 5 39.38 7.2 33.307 9.6 23.98 8.5 35.079 10.7 29.389 10.2 19.364 13.7 To make the relationship more obvious we made a line graph of the two, as shown in Figure 3. Figure 3. Relationship between sales volume and average unit price of cauliflower category Through the line graph we can roughly see that there is a certain relationship between the two, so a regression model is established for correlation analysis: 𝑦 _ 𝛽 𝛽𝑥 _ 𝜉~𝑁 0, 𝜎 (2) Eq.𝜉 obeys a normal distribution, with𝛽 is the constant term coefficient and β is the coefficient of the independent variable 𝑦 _ denotes the price of cauliflower on the first 𝑖 day cauliflower pricing 𝑥 _ denotes the price of cauliflower on day 𝑖 We imported the obtained data into MATLAB and calculated the Pearson’s correlation coefficient of the two and finally came up with the result of R=0.5009. The overall regression coefficient is not zero i.e. there is a linear relationship between the variables. Significant 𝑝 is 0.0004, which presents significance at the level, so the model basically meets the requirements finally the relationship is 0 10 20 30 40 50 60 70 80 90 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 sales volume Average unit price 355 obtained as follows: 𝑦 _ =10.2865-0.0466𝑥 _ (3) In order to visualize the change between the two more the regression equation was fitted to the original data observed as shown in Figure 4. Figure 4. Regression equation fitted to the original data In this question we mainly use the random forest model to make predictions about the total daily replenishment and pricing strategies for the coming week. The random forest model has the following characteristics:(1) It is able to deal with high-dimensional data and a large number of features without the need for feature selection. (2) Capable of estimating feature importance during the training process to help understand the key features of the data. (3) Able to handle missing data and maintain good performance. (5) Has good robustness and can handle noise and outliers. Based on the total sales volume of each vegetable category from June 5-30, 2023, the total replenishment volume for June 5-30, 2023 will be Based on the Random Forest model to predict the sales volume of each category for the next week, the model of Cauliflower category is used as an example to predict the replenishment volume of the Cauliflower category for the coming week by applying the Random Forest model. According to the random forest model, the total sales of each vegetable category for the next seven days can be obtained, as shown in Table.4. Table 4. Forecasted sales volume cauliflower philodendron capsicum eggplant wholesale edible mushroom Aquatic rhizomes 2023/7/1 16.1077 134.5336 100.8684 89.007 81.1291 19.3111 2023/7/2 16.071 133.3179 101.1612 91.3433 82.0071 19.4101 2023/7/3 16.0344 132.1021 101.454 93.6797 82.8851 19.5091 2023/7/4 15.9978 130.8863 101.7469 96.0161 83.7631 19.6081 2023/7/5 15.9612 129.6706 102.0397 98.3524 84.6411 19.7071 2023/7/6 15.9245 128.4548 102.3325 100.6888 85.5191 19.8061 2023/7/7 15.8879 127.239 102.6254 103.0252 86.3971 19.9051 Based on the ratio of wholesale price to sales price from July 1-7, 2023, the pricing strategy for 1-7, 2023 is derived, and the pricing strategy for the next seven days is calculated by the random forest model, as shown in Table.5. Table 5. Forecast Pricing Strategy cauliflower philodendron capsicum eggplant wholesale edible mushroom Aquatic rhizomes 2023/7/1 0.6212 0.6089 0.5086 0.5247 0.4503 0.4751 2023/7/2 0.6286 0.594 0.4992 0.5032 0.4256 0.4334 2023/7/3 0.636 0.5791 0.4899 0.4816 0.4009 0.3917 2023/7/4 0.6433 0.5643 0.4805 0.4601 0.3762 0.3499 2023/7/5 0.6507 0.5494 0.4712 0.4385 0.3515 0.3082 2023/7/6 0.6581 0.5346 0.4619 0.4169 0.3268 0.2665 2023/7/7 0.6665 0.5197 0.4525 0.3954 0.3021 0.2247 4. Individual Product Replenishment Strategy Development Screening of a supermarket's measured data to obtain 36 saleable varieties that meet the requirements, and then through the profit formula and the loss rate formula comprehensive analysis of these 36 varieties, and finally determine the 33 saleable varieties sold on July 1st. By obtaining the replenishment volume and pricing strategy of individual items on July 1, the 33 sellable varieties are finally selected to maximize the revenue of the superstore. The predicted single item replenishment for July 1 is obtained, as shown in Table.6. Table 6. Individual product replenishment on July 1 cauliflower philodendron capsicum eggplant wholesale edible mushroom Aquatic rhizomes 16.1077 134.5336 100.8684 89.007 81.1291 19.3111 By filtering the data, we can come up with 36 sellable varieties that fit the question , then calculate the profit and 356 attrition rate for these 36 sellable varieties and select 33 sellable individual items with the following profit formula: 𝑙 𝑝 𝑗 𝑥 (4) where 𝑙 represents the profit of a single product, and p represents the wholesale price of a single product, and 𝑗 represents the purchase price of a single product, and x represents the sales volume of a single product. Profit formula is as follows: Β 1，𝑆 9.43% 2, 𝑆 9.43% (5) where the wastage rate is S, the average wastage rate of the sample is𝑆 , B is the corrected shelf life,where the wastage rate S 0 B is the modified shelf life, where the loss rate S. The loss rate S can be obtained from annex IV, the loss rate of annex IV can be averaged to obtain the average loss rate of the sample, where B can be regarded as the corrected loss coefficient. The final results obtained see excel third ask process, after calculation, due to the small wrinkled skin (copies), green peduncle loose flowers, zhijiang green peduncle loose flowers of three types of single product profit is low, so it will be rounded off, rounded off after the remaining 33 single product of the total profit of 812.436. Of these, cauliflower, eggplant, edible mushrooms, chili peppers, and foliage replenishments have not yet reached the total replenishment, and by scaling up the number of this single category in equal parts, we get a total profit of 1,175.539. By collecting and analyzing various factors that affect the replenishment and pricing decisions of vegetable commodities. For example, data-driven decision-making, i.e., based on data analysis of various time periods, enables superstores to better understand the sales of vegetable commodities, market demand, etc., and make more informed replenishment and pricing decisions. Seasonal demand, i.e. different changes in sales and demand for vegetable commodities in different seasons. For example, some vegetables may be in short supply in a particular season, while they may be stagnant in other seasons. Costs and profits. When formulating pricing strategies, supermarkets assess the degree of market competition and price elasticity of vegetable commodities by taking into account production costs, storage costs and transportation costs, etc., to ensure that pricing covers costs and maintains a reasonable level of profit. Focusing on customer feedback and market trends, i.e. the superstore collects consumer feedback on the quality, selling price and taste of vegetable commodities, as well as competitors' sales strategies, so as to adjust the replenishment plan and pricing strategy of vegetable commodities. Through these uncertainties, the superstore collects and analyzes relevant data, establishes relevant models, and gives the corresponding replenishment and pricing decisions for vegetable commodities according to the future development trend. This enables supermarkets to have a more comprehensive understanding of market demand, inventory, market competition, and supply chain operations, so that they can more accurately formulate replenishment and pricing strategies for vegetable commodities. 5. Conclusion This paper takes the measured data of a superstore as an example to formulate a set of effective pricing and replenishment decisions for vegetable products, and the comprehensive consideration ensures the balance of supply and demand, and reduces the loss of superstores and the loss rate of commodities. First of all, the sales of each category and single product in different time periods were counted, and the Pearson correlation coefficient was calculated to get the correlation intensity level and interrelationship between the representative eggplant single product, and the distribution law of the sales of each category and single product of vegetables was obtained by organizing and analyzing the data, so as to know the sales of each category of vegetables in different time periods, and the sales of different single products in the same category in different time periods. Popularity. Then, analyze the relationship between the total sales volume of each vegetable category and the cost-plus pricing, establish a regression model based on the measured data, and the results show that there is a regression relationship between the total sales volume and the cost-plus pricing, based on which, the pricing strategy is determined by the positive correlation between the ratio of the wholesale unit price and the sales unit price and the establishment of the Random Forest model to predict the total replenishment volume of the next week, and arrive at the pricing strategy of each category, and again Random Forest model is used to derive the pricing strategy for the next seven days. Finally, obtaining the replenishment volume and pricing strategy of individual items to maximize the revenue of the superstore, these individual items are scaled up in equal proportions, and the total profit can be calculated to be $1,175.539. The model established in the paper basically solves the given problem, combining strong practical strength and high computational efficiency. References [1] Huang, Jianxing. Research on Vegetable Price Fluctuation and Forecasting. South China Agricultural University, 2019. [2] PENG Hongxing, ZHENG Kaihang, HUANG Guobin et al. Vegetable price prediction based on BP, LSTM and ARIMA models. Chinese Journal of Agricultural Mechanical Chemistry, 2020, 41(4): 193-199. [3] Yan Zhengxu, Qin Chao, Song Gang. Random forest model stock price prediction based on Pearson feature selection. Computer Engineering and Applications, 2021, 57(15): 286- 296. [4] Cui, Yun-ho. Research on Fresh Vegetable Sales Prediction Based on CNN-PSO-LSTM Combined Model. Anhui Agricultural University, 2022. [5] Lv Bin. Research on Vegetable Supply Chain Integration. Fujian Agriculture and Forestry University, 2010. [6] Han, W.-G. Research on inventory management program of frozen products category in community vegetable direct stores in Urumqi. Xinjiang Agricultural University, 2017.