Research On Fresh Agricultural Products Distribution Based On Data Mining And Pre-warehouse

With the development of society and the continuous development of economy,people are more and more concerned about the convenience and quality of post consumer service.Fresh electric business is also developing rapidly.COVID-19 has reached its peak in 2020.Meanwhile,the No.1 central document has also repeatedly proposed to solve the problems of agriculture,rural areas and farmers,namely,farmers' income and efficient circulation of agricultural products.Fresh agricultural products are different from other goods.They are perishable and have short shelf life.In the process of transportation,they need high-cost cold chain transportation.At the same time,fresh agricultural products also have the attributes of low single product value and low gross profit.Therefore,this paper studies the terminal distribution of fresh agricultural products,solves the problem of high distribution cost,improves the distribution service quality of fresh agricultural products,gets closer to consumers,improves the sales volume of fresh agricultural products and the income of farmers,solves the upward problem of agricultural products,and targeted poverty alleviation.This paper summarizes the fresh agricultural products distribution and the application of data mining technology in the logistics industry at home and abroad,the related concepts of fresh agricultural products distribution and China's national conditions,and discusses the necessity of fresh agricultural products distribution research.According to the situation of terminal distribution in urban area,a set of solutions is designed,including the location and distribution area division of front warehouse,the selection of front warehouse distribution mode and the task allocation of "front warehouse+crowdsourcing" mode.In the first step of the scheme,the demand point data is collected by using the scrapy framework,and the longitude and latitude are converted into distance units and then input into the clustering model to obtain the location of the front warehouse and the division of the distribution area,and the actual case is analyzed.In the second step,after locating the position of the front warehouse,the warehouse allocation mode of the front warehouse is classified and selected by using the data mining algorithm.By inputting the collected nine relevant feature data into different algorithm models(xgboost,SVM,logistic regression,decision tree),adjusting the model parameters,and comparing the AUC value(the index used to compare the performance of different learners),the xgboost algorithm model with AUC value of 93%is finally selected,and the case data is predicted.Finally,this paper analyzes the distribution mode of"front warehouse+crowdsourcing",and transforms the problem into a weighted bipartite graph model.The weight takes into account the willingness of crowdsourcing delivery workers and consumers' concerns about crowdsourcing delivery services,which is composed of the reputation value of crowdsourcing delivery workers and the distance of normalized distribution tasks.The model is solved by KM algorithm,and a case study is carried out to solve the task allocation problem reasonably.