Research On Correlation Coefficient Of User Portrait Acquisition Based On Machine Learning

With the rapid development of communication technology and the popularization of intelligent terminals,the traditional profit model has gradually changed.With the growth of mobile users,telecom operators have accumulated huge amounts of data.For operators,if they want to open up new markets and win innovative competition,they must effectively use their own data advantages to transform into Digital operators.Analyzing user behavior and personalizing accurate recommendation can help operators to gain benefits.The work done in this paper is as follows: using the real business of the database of X city operators in 2015,including identity attribute information,communication side consumption information,business information,data understanding,data conversion,data cleaning and other work on the data set,using Python language programming,call decision tree,logical regression,support to the direction The decision tree model is more suitable to study whether the user changes the machine or not.Using ensemble learning method to improve the model,the XGboost algorithm has the best classification effect.Kmeans clustering algorithm is used to classify users into three consumption levels: high,medium and low,and Apriori algorithm is used to analyze the association degree of the types of websites that users are interested in.The results show that the users of educational,social and factual websites also pay attention to the type of IT websites.This experiment starts from the actual demand,and applies machine learning algorithm to the study of the relevant indicators of user portraits.It plays an important role in analyzing the needs of the market staff and improving the success rate of sales.There are still some imperfections in this study: such as data processing,model upgrade,clustering methods and other issues,in the follow-up study to improve.