Prediction and Analysis of Shoes Popularity Based on Random Forest Regression Model

Abstract

Since the era of big data, e-commerce has become a popular industry. People are more inclined to online shopping websites and apps, which are more convenient ways to buy commodities. In order to increase the popularity and fame of websites and apps to attract more customers, e-commerce often analyzes and models customer feedback and purchase types in order to get customers' higher preferences for the performance of products. This paper collected a variety of popular shoes scoring situation, current price, sales volume, wanted quantity, releasing date, releasing price and other relevant quantities. The HM aggregation operator is used to define the popularity (the popularity of the shoe). In this paper, the random forest algorithm is used to conduct regression modeling analysis with the pre-processed data, and the importance estimation of each criterion and a regression prediction model are obtained according to the established model. Then this paper analyzes the shortcomings of the model, and adds data normalization processing to the steps of data preprocessing to make the model more powerful in estimating outliers. According to the results, some suggestions are made for the research and development priorities of footwear companies and it is verified that the model can basically complete the prediction of popularity.