Personal Credit Risk Assessment Based On P2P Network Lending

P2P network lending is a business model that gathers funds to lend to the people who need funds by internet,its target consumer group are the ones that traditional finance cannot reach.In 2007,China's first P2 P network lending platform PaipaiDai was established in Shanghai.Since 2013,internet lending has gained rapid development with the rise of internet.In 2016,the number of online credit users in China has reached 160 million and P2 P transactions reached 14955.1 billion yuan.However,the geometric growth of the number of P2 P network lending platforms does not represent the maturity of the P2 P network lending platform.Being a credit business as well as the generally small scale of P2 P network lending companies,their risk management capabilities are far lower than commercial banks,While the biggest risk lies in credit risk.Personal credit assessment can essentially be regarded as a two-category problem.This article takes the personal credit evaluation method as the research object and the lending club website data as an example,while ordinary under_sampling method will result in loss of data information,this article rises a new method which combines the boosting and support vector machines to deal with the unbalanced data,After the unbalanced samples are processed by the arising method,the recall rate and precision rate of the model are improved.Then Gradient Boosting Decision Tree,Random Forest,SVM and Logistic Regression are applied to make predictions.It was found that the Logistic Regression is more robust to imbalanced datasets than the other methods,While Random Forest performs best.