Credit Risk Assessment of Commercial Banks using Pelican Optimization Algorithm based Long Short-Term Memory with Gated Recurrent Unit

With the enhancing tendency of economic globalization, China's pace of sponsoring interest rate marketization is quickening progressively and financial market instability is enhancing every day. The credit is a most important element in bank and financial industries and it is described as unpredicted events that is majorly happens in a form of assets. The Pelican Optimization Algorithm based Long Short-Term Memory with Gated Recurrent Unit (POA-LSTM-GRU) is proposed in this work which is used to assess the credit risk of commercial bank by using collected data. The dataset is preprocessed by Z-score normalization which is used to remove data outliers to enhances the data quality. Then, POA is used for feature selection which selects relevant features and LSTM-GRU is used for classification which learn to capture long-term dependencies. The precision, fl-score, recall and accuracy are taken for evaluating POA-LSTM-GRU performance. The POA-LSTM-GRU attains precision 99.08%, fl-score 96.81 %, recall 97.35% and accuracy 98.76% when compared to existing technique such as Genetic Algorithm Neural Network (GANN), and Self-organizing LSTM.