An Advancing Financial Credit Risk Forecasting Model Using Graph Convolutional Networks for Sustainable Economic Analysis

Credit risk management is essential for financial stability in lending organizations. It involves evaluating the probability that borrowers will fail to repay their debts, which could lead to substantial losses for the institution. Accurate credit risk forecasting is crucial for safeguarding institutions from defaults and maximizing returns. Conventional statistical methods, though effective, often fail to capture complex, nonlinear relationships among variables, resulting in prediction errors in diverse credit profiles. The advent of Artificial Intelligence (AI), particularly Deep Learning (DL), in credit risk management signifies a key progression in addressing these drawbacks. AI, especially DL, enables processing of extensive data and the extraction of significant insights to enhance predictions. This paper presents an Advancing Financial Credit Risk Forecasting Model using the Graph Convolutional Network (AFCRFM-GCN) technique. The aim is to develop a robust and intelligent framework for accurate credit risk prediction to support sustainable economic analysis. In the data preprocessing stage, the min–max scaling method is used to normalize the financial data. Furthermore, the Pelican Optimization Algorithm (POA) is employed in the Feature Selection (FS) process. Moreover, the Graph Convolutional Network (GCN) is utilized for credit risk classification. Finally, the Levy Flight-based Red Fox Optimization (LFRFO) is implemented for parameter tuning. The comparison study illustrates a superior accuracy value of 98.56% over existing models on the Credit Risk dataset.