Credit risk analysis in agri-supply chain finance: genetic algorithm & neural network model

Abstract

The present study analyzes the various risk assessment methods employed in the agricultural supply chain finance (SCF) industry to mitigate its associated credit risks. A backpropagation neural network (BPNN) is trained using a genetic algorithm (GA) to calculate the initial weights and thresholds before evaluating credit risks. This is undertaken in response to the difficulty of choosing the characteristics and the many elements that affect credit risks. Applying the case analysis technique to verify the suggested methodology establishes the most effective credit risk assessment strategy. The results demonstrate that GA-BPNN enhances the rate at which BPNN converges and mitigates its drawback of quickly becoming trapped in the local minimum. Using the PCA technique, we can simplify the process of choosing evaluation indicators and promptly discover representative indicators for assessing agricultural credit risk. The verification findings demonstrate that the GA-BPNN method improves the speed and accuracy of credit risk prediction for agricultural SCF. Financial credit risk estimation can be used to assess the accuracy of GA-BPNN in predicting risks related to financial resources in supply chain financing of agricultural enterprises, hence lowering credit risks in agricultural supply chain financing.