| With the advent of the era of artificial intelligence,the enterprise is faced with various cost challenges,although the modern management of the catering industry has become more mature,but also due to a surplus of food,causing food rotting waste,thereby reducing profits,or food is in short supply,unable to meet consumer demand,leading to turnover fell,the customer satisfaction levels drop.Nowadays,many supermarkets use artificial prediction,and sales staff judge the number of vegetables to be purchased the next day according to their own experience.Because the sale of vegetables is affected by a variety of factors,the accuracy of artificial objective judgment is very low,which leads to the phenomenon of insufficient supply of some dishes and shortage of goods.Therefore,if enterprises want to stand out in the market,they must gradually develop from the traditional sales management mode to the informationization and automation management mode.In order to solve the problems of food supply shortage and food waste in vegetable sales forecasting,this paper proposes a vegetable sales forecasting method based on ensemble learning.LightGBM model is first chosen,length of LSTM memory network model of two single prediction model for research,because the sales forecast is affected by the linear and nonlinear,then put the two models are combined,thus improve the prediction accuracy,the combination model of arithmetic average method of solving the weight combination prediction,entropy weight method,the variance reciprocal method,integrated learning stacking method fusion model comparison.The experimental results show that the combined model compared with the single model has higher prediction accuracy,in the combination model,based on the integrated study of stacking method has higher accuracy of prediction,this article proposed based on integrated learning method in the middle of vegetables prediction accuracy reached 94.19%,the highest 92.37%,followed by roots like,melon and solanaceous fruit 91.95%,the lowest is 86.63% fresh beans and fungi,as a whole,this method can effectively predict the vegetables in the short term sales in the future.Finally,this paper designed and realized the Web end system of vegetable sales forecast,using the visual interface to facilitate the view of the future short-term and historical sales of some vegetables,so as to make procurement plans for enterprises,enhance the vitality of enterprises,so that enterprises can get better development in the market. |
