Research On 95598 User Portrait And Behavior Evaluation Based On Power Big Data

At present,China Southern Power Grid has clearly put forward the strategic goal of "creating a first-class international power grid enterprise with lean management,fine service,excellent performance and excellent brand" in its overall reform and development plan for the 13 th Five-Year Plan,as well as the corporate philosophy of "honesty,service,harmony and innovation" and the service philosophy of "customer first,harmony and win-win".With the advent of the big data era,the volume of 95598 user data in the power industry is increasing explosively.Due to the lack of evaluation and intelligent analysis of internal and external factors affecting user behavior by power companies,it is impossible to measure the impact degree and duration of chemical single data and user complaints,which seriously hinders various performance evaluations of companies and employees.Therefore,reducing customer complaints,improving the limitations of current power company service analysis,and improving the level of forecasting power customer behavior have become one of the research hotspots of power companies.The main work of this paper is as follows:Firstly,this paper takes 95598 users in Guangxi in the first quarter of 2018 as the research object,and introduces the concept of user portrait into the field of user behavior evaluation.Through big data technology,95598 users are deeply mined to obtain rich information hidden behind the users.This paper deals with big data from the aspects of data segmentation,checking and cleaning,and constructs a multi-dimensional user portrait model,which also lays a foundation for the research of user behavior evaluation model in big data environment.Secondly,on the basis of 95598 user portraits,this paper combines machine learning techniques such as random forest and decision tree to analyze and screen user variables,find out the important variables that affect user behavior,deeply analyze the causes of user complaints,and help power companies find out the weaknesses of the power grid.At the same time,a series of sub-models with user behavior evaluation capability are generated,and the output results of the sub-models are fused with Logisitcs regression model to construct an R-L model for power user behavior evaluation.Finally,this paper randomly selects some data of power users as samples to verify and analyze the model.At the same time,the R-L model and the traditional Logisitcs model are tested respectively in terms of distinguishing ability and stability,and the user behavior evaluation effects of the above two models are compared and analyzed through examples.The final results show that the R-L model has high accuracy in predicting the user’s behavior.The model can also help power companies to provide professional market education pilot services to users and avoid potential complaint risks,thus continuously and effectively meeting users’ needs and achieving the effect of improving users’ service quality.It has important practical significance.