Prediction Of Advertising Click-through Rate Based On Combination Of XGBoost And LR

The Internet has gradually become the main way for people to obtain information and entertainment.In the Internet industry,Internet advertising has become the most important profit model,monetization has become a very important evaluation standard in Internet products.In order to measure the effect of Internet advertisement,the click-through rate of advertisement is introduced as an important index.The high click-through rate of advertisement indicates that the effect of Internet advertisement is good.As a result,ad clickthrough estimates are an important task,and accurate estimates can improve the effectiveness of your advertisement,helping advertisers find the right audience and deliver the right advertisement in the right media.The logistic regression model is the simplest and most widely used model in industry,using the logit transform to map function values to the 0~1 interval,and the mapped function value is the pre-valuation of the ad click-through rate.The XGBoost model is all called the extreme gradient lifting tree,and the XGBoost model is the extension of the gradient tree boosting decision algorithm,based on all the trees generated by the gradient lifting algorithm in one step above,moving in the direction of minimizing the given objective function.XGBoost model is a novel tree learning algorithm for processing sparse data.In addition,the theoretically reasonable distributed weighted histogram algorithm makes it possible to deal with instance weights in approximate tree learning.This thesis mainly uses the XGBoost model to train the data,find out the feature combination way,and then use the logistic regression model to predict.The paper first introduces the definition of the process of predicting advertising clickthrough rate,and gives the main differences between search and display advertisement.Secondly,the advertising click-through rate Prediction model is introduced,including the logical regression model,the XGBoost model and the LR model and the XGBoost model of the combination of the way and algorithm.Then,in order to verify the validity of the model of the combination of LR and XGBoost,the experiment mainly includes preprocessing the data,analyzing the variables in the dataset,processing the unbalanced samples,training the model and determining the optimal solution of the model parameters.Finally,different models are used to predict click-through rate and compare their performance.