Risk Assessment And Early Warning Analysis Of P2P Network Credit Platform In China

P2P network loan originated from private lending,driven by the Internet technology,P2 P network loan began in some developed countries to develop rapidly.2007 P2 P network loan into China,and soon in China this land grow up.After years of exploration and practice,P2 P network loan has been gradually recognized by all walks of life,and attract a large number of users.However,China's credit system construction is lagging behind the Western countries,the relevant regulatory regulations are still not perfect,and the platform's own wind control capacity and internal control capacity is weak.Therefore,China's P2 P network loan industry,although the momentum of rapid development,but the platform Paolu,closed,difficult to mention such problems arise.Firstly,the paper uses the data mining theory and risk assessment as the foothold,draws on relevant literature and books at home and abroad,and constructs the risk evaluation index system of P2 P network loan platform in four aspects: operational capability,development potential,platform quality and profitability.Due to the low number of high-risk platform samples and the serious imbalance in the sample,the SMOTE algorithm is used to balance the sample data,making the high-risk platform and non-high-risk platforms account for roughly the same.Then,three kinds of data mining methods,logistic regression,decision tree and support vector machine are used to establish the risk assessment model respectively.The ten-fold cross validation and confounding matrix method are used to calculate the average of the ten operations as the accuracy of the three evaluation models evaluation of.The empirical results show that the accuracy of each evaluation model is from high to low: CART decision tree> RBF support vector machine> Logistic regression,the overall view of CART decision tree model is more robust.In the process of model establishment,this paper uses the real data of the platform to predict the test set samples.The prediction accuracy of the three models is high,especially the CART decision tree model.The evaluation results can provide reference for the field of risk assessment.Next,the platform is not yet a serious problem of early warning risk research,based on a comprehensive index,operability,sensitivity and dynamic continuity,to build risk warning indicator system net loan platform.Combining the principal component analysis with the improved KLR signal analysis method,277 platforms are selected as the samples,and the collected data is collected,processed and substituted into the model.The risk ranking and risk warning status of each platformare obtained.Then the warning results are substituted into the BP neural network for training,and the test set is predicted.The empirical results show that the prediction results of the training model are better and the forecasting model has certain application value,and it can be concluded that the early warning results are reasonable.The focus and difficulty of this paper is to apply the data mining method to the risk assessment of P2 P network loan platform,construct the risk evaluation system,and evaluate and forecast the P2 P network loan platform.Based on the above empirical results and conclusions,this paper puts forward some suggestions on the healthy and sustainable development of P2 P network loan industry in China:(1)To strengthen the supervision of network loan industry and increase the transparency of network loan industry(2)To gradually improve the extensive personal credit system,The establishment of network loan industry integrated database(3)The establishment of P2 P network loan risk early warning management system,the risk of early warning legislation.