The credit risk assessment of P2P lending based on BP neural network

P2P lending is a new type of financial model under current Internet financial circumstance. The evaluation of P2P Internet loan has tremendous meaning since its credit risk is the leading factor of Internet finance stability. According to the characteristics of P2P loan, BP neural network model was introduced to evaluate the risk of credits quantitatively. We build an empirical model with the lending club data. Given the experimental result, P2P Internet loan credit risks are primarily determined by several key attributes. In addition, BP neural network method offers an evaluation result with 78.6% accuracy rate (type one falsification rate 4.8% while type two falsification rate 16.6% respectively), which indicates that risk evaluation method has a predominant performance.