Study On P2P Personal Credit Scoring Models

With the development of internet,Peer to Peer industry is rapidly developed in our country.Compare with Peer to Peer industry in western country,we have some disadvantaged that we have not complete individual credit reporting system.So that this company should be build their individual credit reporting systerm by their client.Many scholars used to do the research about credit scoring models through the data of bank.But the data between bank and Peer to Peer have something difference.In this paper,I use the data of Peer to Peer and some models about the statistics and machine learning to systematic show the process that from preprocess to building model and evaluating model.When we build the individual credit scoring model,have some traditional method.This method can be divided into statistical method and non-statistical method.Statistical method : discriminate analysis,logistic regression and decision tree etc.Non-statistical method :linear programming,genetic algorithm and neural network.In this paper,I compare with logistic regression,KNN, neural network,random forest.Finally,we find that random forest have excellent performance in those method.