Design And Implementation Of The Processing And Analysis System Of Credit Rating Based On Data

Credit can reflect the degree of personal integrity as a good quality in our society.We secure the trade in our trading platform through the credit.You will get more assurance amount and more rights in our system with better credit rating.At the same time,it provide a reference to help buyers and sellers judge the risk in the trade and help complete trade security.This paper is based on the actual project of the international station of Alibaba B2B-- the trade assurance system what completes a series of data driven, user credit rating of the processing and analysis work.The system includes data acquisition,data preprocessing, data analysis and monitoring, model selection and training,system test and model evaluation.Data acquisition module accomplish the work of data acquisition of different kinds of data sources in several ways.Data preprocessing module includes data cleaning, data construction,data normalization,data formatting,data verification and so on. Data analysis and monitoring module includes credit analysis work based on rule engine and real-time data monitoring.Data monitoring module secure the real-time priority of data and realize the function of data feedback.Model selection and training module includes model selection and model training work.In this module,we analyze the application scene and the advantages and disadvantages in several classification models.Then we choose the models what fit our requirement and describe the details of model training.System testing and model evaluations includes unit testing, module testing and model evaluations.Model evaluations module evaluates the classification models considering training time,accuracy rate,the size of training size and so on.