Research On User Mining Of D Company's Wealth Management Products Based On Improved RFM Model

In the era of mobile Internet,in addition to pulling new users,the most important means of obtaining traffic for various products is to improve conversion,retention and promotion.The era of Internet dividends has passed,and traffic costs are getting higher and higher.At this time,it is necessary to spend the least operating costs to operate users and guarantee revenue.Through the refined operation mode,the problem of user operation is solved through dataization,and only the refined operation can retain users.In this thesis,based on the improved RFM model,k-means clustering of users of D enterprise wealth management products is carried out to realize value stratification.After using R language to obtain results,it is found that users are divided into five layers,which are respectively important development users,important to maintain users,important retention users,general value users and general development users.Then use Boruta and PCA algorithn to screen the important characteristics of each user group,select more suitable models through experimental results and performance comparison,and understand the consumption attributes,financial attributes and basic attributes of each type of users,so as to Class users design corresponding marketing programs.The innovation of this thesis is to combine the RFM model with the feature selection model for the research of user mining problems,and to improve the RFM model,and to increase the repayment status indicators for the overdraft consumption function of Internet financial products.The user data is divided into three categories:behavior,basic attributes and preferences.According to the nature of the data,it is analyzed and processed.The application of Boruta and PCA is verified.In general,many commonly used algorithms are used.Features advantages to solve real-world user operations problems.At the end of the article,we observe the data after the program is put into use for one month,and find that the conversion rates have been improved.It proves that the RPM model and feature selection model used in this thesis are still useful and effective in actual operation,in the overall operation system.It is used as an effective analysis tool to help the actual operation of the company's products.