Abstract This study applies machine-learning techniques and rulebased methods to on struct nonlinear nonparametric models to forecast retail consumer and medium-sized enterprises (SMEs) credit risk. By combining customer transactions and enterprise data from 2018 to 2020 sampled from a major business district in the People’s Republic of China, forecasts were constructed that significantly improved the classification rates of customer and enterprise delinquencies and defaults. Moreover, the time-series patterns of the estimated delinquency rates and credit scores over multiple dimensions produced by this model suggest that aggregated credit risk analytics may have important applications in forecasting systemic risk, which might shed some light on obtaining prospective insights regarding consumer credit that can be gleaned from historical data especially pandemic period.

Keywords: Credit risk model, Machine Learning, Rule-based,Credit Score