<![CDATA[Banking stability, particularly the risk of financial distress in private commercial banks, remains a critical issue that requires accurate and reliable prediction models. This study aims to analyze the characteristics of financial distress in Indonesian private commercial banks and to evaluate the effectiveness of Artificial Neural Networks (ANN) and ANN optimized with Particle Swarm Optimization (ANN-PSO) in predicting financial distress. Using financial data from 59 private commercial banks over the 2020–2023 period, this research employs five financial ratios as input variables and applies ANN and ANN-PSO models, with parameter selection conducted through a trial-and-error and optimization process. The results show that financial distress peaked in 2022–2023 with 32 distressed banks, while descriptive statistics indicate differences between distress and non-distress banks, including average NPLs of 1.40% versus 1.04%, ROA of 0.36% versus 0.75%, and LDR of 93.89% versus 92.39%, respectively. In predictive performance, both ANN and ANN-PSO achieved identical test accuracy of 95.74%, sensitivity of 93.75%, specificity of 96.77%, and an F1 score of 93.75%, although ANN-PSO demonstrated better model stability with lower training accuracy (98.40%) compared to ANN (99.47%), indicating reduced overfitting. Despite these promising results, this study is limited to a relatively short observation period and a fixed set of financial ratios; therefore, future research is recommended to incorporate longer time horizons, additional macroeconomic variables, and alternative optimization techniques to further enhance prediction robustness and generalizability.]]>
Copyrights © 2025
