A Study On The Method Of Click-through Rate Prediction Based On Machine Learning

In recent years,with the rapid development of computer technology and Internet technology,network marketing with massive data as background has been gradually favored by all fields.Then,as a new form of advertising,online advertising should be born,which shows great,potential commercial value and market competitive strength.At the same time,online advertisers have large scale advertising search information,so they can provide advertisements that meet users’ needs according to users’ search history,and this has also become an important income channel for the Internet industry at present.For search advertising,the prediction of advertisement click through rate is the most fundamental and key technology.The technology not only directly affects the emission location of advertising in the Internet,but also involves the specific charging state of advertising clicks.Thus,it can be seen how to predict the ad click rate based on machine learning and making full use of the user’s search historical data as a significant practical work.Based on machine learning,we can not only predict log data in advertising system,but also calculate the probability of users clicking on candidate advertisements,so as to ensure that advertisements with larger click probability are displayed to corresponding users.In this study,based on machine learning CTR prediction method research subjects were analyzed by a large number of references to collect,understand the results and the present situation of domestic and foreign scholars in the CTR prediction research are obtained,and on the basis of the research background,research purpose,research significance.After that,this study explores the concepts and technical foundations related to machine learning based advertising click through rate prediction,including the concept of computational advertising,the technological foundation of shallow machine learning model and deep machine learning model,and the concept of AUC index.The machine learning model CTR prediction,this paper from the Naive Bayesian model,logistic regression model,support vector regression model analysis of three aspects of shallow machine learning model,and the artificial neuron model,BP neural network,the neural network model of the depth of the three aspects of the depth of machine learning model.Then,this research further explores the depth neural network model of the prediction of the advertising click rate.In order to ensure the prediction method of advertisement click through rate based on machine learning is scientific and practical,this study evaluated the prediction method by experiment.In view of this,this study set up experiments according to the demand of machine learning click through rate prediction,and summarized the application advantages of machine learning based advertisement click rate prediction method based on the results of shallow learning model test and deep learning model test.