Research on P2P Credit Assessment Based on Random Forest ― from the Perspective of Lender’s Profit

Without the guarantee of traditional intermediary, the benefit of lenders is often damaged by the default of borrowers. Now, many machine learning models have been widely used on credit assessment in P2P online lending. To further improve the validity of credit assessment and the profit of lenders, the random forest algorithm with the profit of the lender as the primary evaluation criterion is proposed. We choose fifteen features as the input vector of our model, then the profit of lenders is obtained by formula. The dataset from Lending Club was applied to validate the effectiveness of this model. The findings of research show that the interest rate, debt-to-income ratio and loan amount are the three most important factors of default. Furthermore, random forest model can achieve higher lender's profit compared with other traditional methods, so it could be applied to credit process control in reality.