Application Of Data Mining In Improving The Realization Efficiency Of Internet Advertising

In the context of the rapid development of the information Internet era and the mass popularization of mobile devices,people have been gradually changing their way of obtaining information from the traditional media such as newspaper,news,and TV to Internet and online,and more and more people access public networks through mobile phones,pads and other mobile devices to obtain relevant information quickly and efficiently.On the basis of the high aggregation of information content and the good stickiness of users.This paper is directed and focused on how to improve the realization efficiency of commercial advertising as much as possible on the premise of established traffic.Regarding the behavior chain from the beginning of accessing the information to the final call as an effective connection conversion,all factors and actions that improve the conversion efficiency are analyzed,which is the behavioral data that needs to be collected.Based on the statistics and analysis of these data,and combined with some tools and methods,the level of interest of users in the current advertising information will be predicated to help get answers for these questions,such as "why do users browse the information?" and "why do users buy it?",so as to complete the"interest model" training for the retained users.Then,the relevant commercial advertisement information is recalled and sorted based on the users’ interest model to provide positive incentives to the advertising monetization efficiency.This paper will focus on the impact of user behavior on the monetization efficiency of commercial advertising.The user behavior data is defined,collected,mined and analyzed to guide the improvement of the monetization efficiency of Internet commercial advertising based on feature engineering and recommendation platforms.The innovations of this study:Firstly,this study is based on the complete user behavior chain for data collection,subsequent analysis and training,which is different from the analysis idea of single-point behavior or characteristics of users;the integrated analysis of the whole chain behavior is completely consistent with the life cycle of Internet advertising conversion.Secondly,the study of the change rate of CTR by a single feature change is proposed.Thirdly,in the model training,the fusion experiment of the algorithms is carried out,and the result cluster of the basic algorithms is added to the features of the subsequent algorithms for circular training to speed up the convergence of the effect.