Analysis Of User Email Behaviors For Pharmacy Promotion

With Internet+ coming,people's life is much closer with it.The Internet has penetrated into people's daily work and life every day and night.Traditional commination model just like call telephone or writing letter convert to the internet model like sending email or using WeChat to chat.Pharmacy promotion also meet the innovation from Internet +.Running more than 20 years' F2 F sales promotion meets more and more challengers from Internet model,like fast,fresh and customize.And just like other industries,pharmacy is try to use digital multi-channel model to support to traditional work model.Relying on the digital multi-channel promotion not only can increase frequency of the doctors engagements,but also can cover more doctors who haven't been covered by sales,to achieve the widest range of pharmacy promotion.This paper first studies the key technologies of user e-mail behavior analysis in the field of drug promotion.In this paper,we study how to use the ODS model to solve the data quality of data integration,and use the Data Profiling model to discover the abnormal data,thus further ensuring the validity of the data.This paper uses the AUC to evaluate the effect of classification,and selects XGBoost as the classification algorithm by comparing experiments.At the same time,we use the characteristic project to select the e-mail subject,the mail sending time,the doctor's title,provinces and doctors received the number of e-mail and etc.as a classification of the characteristics.On this basis,this paper analyzes the requirements of the user's e-mail behavior analysis system in the field of pharmacy promotion.It analyzes the different data processes in the system,including online and offline.It finds out the key features,and builds use case models.Then it designs system logical architecture using layer style,that is,data integration layer,data validation layer,model analysis layer and presentation layer.It implements the forecasting model with Python,adopts HDFS as the data storage platform,and uses Spotfire to realize the presentation layer.Then the detailed design and implementation of the system layers are described in detail by using the activity diagram,class diagram and timing diagram.Finally,the AUC validity testing and performance testing are carried out in this paper.The test results show that the system has practical reference value for the forecasting of e-mail invoicing rate,and the actual business optimization has achieved the desired effect.