Research On E-commerce User Classification Model Based On Machine Learning

With the development of short video and live delivery,online shopping has attracted a large number of shopping people with the advantages of convenient shopping,thoughtful service and high efficiency.At the same time,e-commerce platforms are becoming more and more,users have more choices in online shopping,and the competition between e-commerce is more intense than before.Therefore,precision marketing for different types of ecommerce users has become a hot topic in the field of e-commerce research.At present,many machine learning models are applied to the field of e-commerce user behavior analysis.However,most machine learning models are optimized by grid search algorithm,and the operation efficiency of the model is low.Therefore,this paper use the machine learning model based on Bayesian optimization algorithm to classify e-commerce users.The model runs fast and can achieve good classification results.The research contents of this paper are as follows : Firstly,based on the shopping data of e-commerce platform users collected by big data websites,exploratory analysis is carried out on the numerical variables,character variables and categorical variables in the data,and the distribution rules of these variables are analyzed.Observe whether there is a positive and negative sample imbalance,and if there is a sample imbalance,data sampling methods need to be used for processing.Then clean the data and convert the data format to prepare for the subsequent construction of the machine learning model.Next,feature selection is performed on the features in the data.According to the selected features,four machine learning models based on Bayesian optimization algorithm,KNN,random forest,XGboost and Light GBM,are constructed to classify e-commerce users and compare the classification effects of several models.It can be seen from the comparison results that among the several models constructed,the XGboost model has the best classification effect,and the overall accuracy of the model reaches 0.97,which is most suitable for constructing the e-commerce user classification model.Further,in practice,the XGboost model is used to output the average value of each attribute of the two types of users in the classification results,analyze the characteristics of high-quality users and non-high-quality users,and output the feature importance in the XGboost model.Analyze the factors that affect the classification of ecommerce users.Finally,according to the characteristics of the two types of users,suggestions are made for the e-commerce platform to formulate sales strategies can help the e-commerce platform maintain high-quality users,and continuously develop non-highquality users to prevent the loss of these users.The research on the classification model of e-commerce users in this paper can provide guidance for the customer development of ecommerce platforms and help the user groups of e-commerce platforms to grow.