Research On Fresh Vegetable Sales Volume Trend Prediction Based On Improved SVM

In recent years,the agricultural products have been the goods,which possess the fastest increasing order volume and the largest quantity on the e-commerce platform,instead of electronic digital brand products.The high frequency and low unit price consumption behavior of buying fresh vegetables online has become the most dynamic and mainstream consumption trend.At present,there are few in-depth studies on the sales volume trend prediction of fresh vegetables at home and abroad.Therefore,this paper takes an ecommerce platform in Anhui Province as the research object,puts forward the sales volume trend prediction method and model of fresh vegetables based on improved SVM,and develops the sales volume trend prediction system of fresh vegetables using Java language,which aims at providing a fresh vegetable sales volume trend prediction model building and application to provide the reference,for the realization of the electricity business management provide guidance and scientific production orders.The main work and achievements of this paper are as follows:1)Improving SVM by proposing particle swarm optimization based on fuzzy information granulation and optimization.Fresh vegetables have the characteristics of seasonality,centralized listing and short shelf life.The sales volume is easily affected by external factors,and the sales series presents the characteristics of periodicity,non-linearity and small sample.According to the sales characteristics of fresh vegetables and the characteristics of sales series,the fresh vegetables sales volume trend prediction method based on fuzzy information granulation and improved particle swarm optimization support vector regression was proposed.2)Based on improved SVM to build the sales volume trend prediction model of fresh vegetables.Combining the advantages of fuzzy information granulation method and optimized particle swarm optimization algorithm to improve SVM prediction,the sorting order data of an e-commerce platform in Anhui Province from January 1,2018 to December31,2019 are predicted.The experiment shows that the built model fitting prediction ability is superior to difference autoregressive moving average(ARIMA),back propagation(BP)neural network and short-and long-term memory(LSTM)model,the selection of 4 kinds of fresh vegetables sales interval prediction mean absolute error(MAE)were 8.78,14.67,14.87,9.55.The validity of the sales volume trend forecasting model of fresh vegetables was verified.3)Research and develop a sales volume trend forecasting system for fresh vegetables.By mining and extracting the sales data of fresh vegetables,the interactive sales volume trend forecasting system of fresh vegetables based on Java Web was developed by combining fuzzy information granular method,optimized particle swarm optimization improved SVM,Spring Boot,Echarts,Bootstrap and other technologies.The sales sequence of fresh vegetables was analyzed and displayed in different window partitioning Settings of the fuzzy information granulation method,and the fitting prediction results of the experimental model were compared and verified.Predicting the sales volume trend of the future of fresh vegetables nearly a period of time can help enterprises to understand the market supply and demand and guide scientific production,reduce unnecessary supplies and circulation to promote the benign development of the science of fresh vegetables industry.