Research On The Sales Of An E-commerce Product Based On Data Mining

In the current e-commerce environment,enterprises are becoming increasingly competitive.To occupy a favorable position in the market requires accurate prediction and judgment of commodity competitiveness.Reasonable prediction will not only bring huge economic benefits,but also maintain market stability development.Combined commodity and sales forecasting are the core content of e-commerce.For example,users may be recommended to purchase commodities to promote the sale of goods based on historical sales orders.Commodity sales forecasts enable companies to make informed business decisions and predict long-term and short-term performance.This article uses e-commerce store sales data and order data as research objects to explore the combined commodity and future sales.The main research contents are as follows:(1)The repurchase rate reflects the users' stickiness of goods or services which is particularly important for enterprises.Exploring the influencing factors of the repurchase rate is of key significance for promoting commodity sales and forecasting sales.(2)Commodity combination has a wide range of applications,such as recommending possible purchases to users based on historical sales orders to promote product sales.Use association rules,method based on association rules and genetic algorithm and improved collaborative filtering algorithm to explore commodity combination.We use stepwise regression to select the variable set and fit the Poisson regression model for each category after clustering.(3)In order to predict the sales of goods,the advantages of the three algorithms of BP neural network,LSTM neural network and Verhulst gray model are combined.For example,BP neural network can combine the current independent variables to predict sales,LSTM neural network can explore the impact of historical data on sales,and the Verhulst model can predict sales based on the growth trend of variables.A BP-LSTM-Verhulst combinatorial neural network based on the AP algorithm is constructed to predict sales.And the prediction accuracy of this method is proved by an example.