Research On Commodity Purchase Plan Of E-commerce Platform Based On BAYES-BP Forecast Algorithm

With the continuous development of e-commerce and the development of consumer needs in the direction of personalization and diversification,for e-commerce platform operators and merchants,to improve operational efficiency and obtain higher profits,it is necessary to enhance product diversity,service capabilities,In-depth discussion on reducing costs and other aspects.This thesis takes the background of an enterprise's self-operated e-commerce platform.In view of the fact that the company's service level on the self-operated platform needs to be improved,the commodity inventory needs to be reduced to improve operating income,and other factors.Considering the long procurement cycle of certain commodities and large inventory occupation To carry out research on procurement forecasting and procurement planning based on the BAYES-BP forecasting algorithm.Accurate procurement forecasting can reduce the scale of inventory for enterprises,increase the flow of funds and reduce the number of purchases,reduce inventory,improve customer service quality,and thus improve the competitiveness of enterprises,which has very important practical significance.Based on the research on the theories and methods used in relevant research at home and abroad,this thesis first analyzes the factors that affect sales in the self-operated e-commerce platform,and finds that the number of product collections,product reviews,product good ratings,etc.The unique factors of the e-commerce platform,these factors have an important impact on the later merchandise sales and purchases.According to Granger causality analysis,the influencing factors are screened.The final selection includes commodity unit price,commodity good rating level,commodity collection number,commodity Nine influencing factors such as the number of reviews,whether the inventory of the goods is sufficient,the type of goods,the number of times the goods are added to the shopping cart,whether it satisfies a week of unreasonable returns,and whether there are coupons.Secondly,construct a prediction model combining Bayesian method and BP neural network,use the BP neural network to build the initial model in the prediction model,use the Bayesian method to optimize the BP neural network,obtain the network hyperparameters,determine At the same time,the optimal weight threshold of the network is proposed to optimize the learning rate.Based on the model construction,the 9 influencing factors are used as the input parameters of the model,and the forecasted purchase demand is used as the output parameter to predict the purchase demand of a certain(category)commodity.Combining inventory parameters to formulate commodity purchase plans.Use the procurement prediction model and the method of formulating the commodity purchase plan proposed in this thesis to apply to the self-operated e-commerce platform for empirical analysis,select 350 kinds of commodity data in June 2019 for analysis and prediction,and compare the prediction results with traditional methods.The comparison shows that the average error and variance of the optimized model are smaller than the traditional method.According to the procurement forecast value generated by the model,the procurement plan is made and it is found that the accurate procurement forecast can effectively reduce the occupation of inventory and excess funds,and has high practical value for the enterprise.The forecasting model and procurement planning method proposed in this thesis can also be extended to other e-commerce platforms,and has a certain reference role.