User Profile And Rating Research Based On Big Data

With the continuous evolution of the Internet and the increasing popularity of smart devices,people can access the Internet in a variety of ways.The Internet has accumulated a large amount of data while bringing convenience to lives.How to mine and analyze these data to play the value of data is a hot topic of research.On the other hand,users play a very important role in the current situation.With the user,enterprise will be income and long-term development.How to analyze the features of users on the basis of big data,so as to provide customized services for different user groups is also a very meaningful research problem.The research contents of this paper is as follows:First,we research the problem of user profile of financial institutions.For a specific financial institution:A city center bank,we use Elkan k-means algorithm to classify users into four user groups based on user's savings data:active user group,potential user group,stable user group,and losing user group.For different user groups,we use user attribute data to image the user group and analyze the characteristics of the user group.Further,based on the user's savings data,the SVR regression prediction algorithm is used to predict the user's savings potential.The prediction accuracy is higher by evaluation at the individual level and the overall level.Secondly,we propose a user rating mechanism.For a specific operator organization:B city operator branch,we extract user data including three dimensions of generating traffic,usage duration,and traffic ratio of different types,thereby obtaining user contribution data and user behavior data.We use Elkan k-means algorithm to divide the user contribution data,and determine the cluster definition according to the degree of contribution of user to operator.So,the user level information is obtained.Based on user behavior data and user level information,we use random forest,CART and logistic regression to predict and analyze.The classification performance of random forest is the best,and the model accuracy is about 8 1%.Finally,we develop a platform for big data visualization and analysis.The platform adopts the MVC architecture,which is mainly composed of two parts:the background and the front end.The background part is built by Django framework,which can realize data saving and reading,data analysis and data transmission.The front-end part is built by HTML5,CSS3 and JavaScript,the chart is presented by ECharts,which can realize dynamic display of data and real-time interaction with users.