Research On Hierarchical Model Of E-Commerce Sellers Based On Data Mining

China's e-commerce has experienced the start-up period,adjustment period,warm-up period and high-speed development period in the past ten years.However,as the demographic dividend of the Internet industry gradually disappears,the e-commerce industry has also entered a period of transformation and upgrading,facing the problem of weak user growth,flat platform performance growth,and high user retention pressure.In this context,many e-commerce companies began to pay attention to the fine operation of the business,which is a good way to solve the above problems.As a means of data mining,dividing the business into different levels can enhance the business operation concept from the perspective of data,and have a very good boost to the refined operation of the business.More and more companies have focused on it because of its great value in both theory and application.The Self Organizing Map netural network has many advantages in self-organizing mapping,good visualization,computational efficiency and clustering effect in practical applications.In addition,neural network algorithms have better predictive effects on data sets with large data volumes.Therefore,we choose SOM algorithm to apply to the construction of the merchant hierarchical model.This paper introduces the SOM algorithm in detail and deeply analyzes the underlying principle of the algorithm.In addition,we redesigned the learning process of learning rate and neighborhood radius to adapt to the experimental environment of this study.After that,we explore the original data set comprehensively based on the R language.At this stage,the default value,outliers,data skew distribution,and dimensional differences have also been resolved.Then we analyze the correlation between the indicators of the data set in detail,and carry out the principal component analysis of the original data set by PCA principal component analysis method.After analysis,14 features are selected to form the input data set of our model.Finally,we use the SOM algorithm to build an hierarchical model,and finally divide the business into three levels:head,waist and tail,combined with the actual business logic.In addition,we also put forward some suggestions for the personalized operation strategies of different levels of merchants,in order to provide a reasonable and effective reference for the business operation strategies of the subsequent e-commerce industry.One of the characteristics of this research is to use PCA combined with SOM algorithm to layer the merchants,and use the means of data mining to assist the personalized operation of the merchants.The paper concludes with a discussion of the limitations of this study and its vision for the future.