Analyzing And Predicting Promotion Effects Of E-commerce Celebrities' Social Behavior

E-commerce celebrities refers to the "internet celebrities" who use e-commerce platform for business activities and carries out marketing with the help of social network platform.The economic scale of the business model based on e-commerce celebrities has reached 58 billion yuan in 2016,surpassing the box office of Chinese movies in that year.To discover and study the typical behavioral patterns of e-commerce celebrities on social network platforms,tap the potential links between e-commerce platforms and social network platforms,so as to deeply understand this new cross-platform economic phenomenon,thus helping to evaluate the commercial value of e-commerce celebrities and improve its social marketing.At present,the research on e-commerce celebrities is still at the level of business cases and qualitative concepts,lacking of quantitative analysis based on scale of actual network measurement.For this reason,this paper takes Sina Weibo and Taobao as specific measurement objects,carries out cross-platform measurement and data fusion,and establishes the original data set of e-commerce celebrities.Based on this data set,it analyses the social business behavior and business activities of e-commerce celebrities on social network platforms and e-commerce platforms,establishes the business value model of e-commerce celebrities,evaluates and predicts the business value of e-commerce celebrities' social behavior.The main work of this paper is as follows:(1)Write web crawler to measure micro-blog and Taobao platform,and form e-commerce celebrities' data sets of social network platform and sales data sets of e-commerce platform.The data set includes 108 e-commerce celebrities' social data on Weibo for 8 years and 4 months,Taobao sales data in April 2018,and the basic information data of 46,470 Weibo users liked and reposted by e-commerce celebrities.(2)This is the first time to construct a characteristic project to depict the marketing behavior of e-commerce celebrities.Based on the social data set of e-commerce celebrities,this paper extracts and analyses the marketing behavior of e-commerce celebrities,including advertising behavior,promotion behavior and word-of-mouth marketing behavior.Mining the statistical rules of the three kinds of behavior,based on the behavior analysis,35 marketing behavior characteristics and 4 daily behavior characteristics of e-commerce red net are constructed,and two basic information characteristics are extracted,a total of 41 social characteristics.(3)For the first time,an evaluation model of e-commerce celebrities bonus sales level based on social features is constructed.On the basis of the above-mentioned feature engineering,a red sales model of e-commerce network is constructed by using random forest,logistic regression,kNN and other classification algorithms,and 10 features that can most affect the sales of e-commerce celebrities are found.The experiment shows that the highest accuracy of the model can reach 0.83.