Aug 2026· Journal of Computer Networks, Architecture and High Performance Computing· Vol 8, pp. 493-503· 0 citations· 15 references
TL;DR
The results show that Logistic Regression proved to be the best-performing model on the held-out test set, and the strongest determinants driving financial resilience were the number of prior delinquencies, loan purpose, and a low debt-to-income ratio, with credit score also contributing.
Abstract
Financial resilience among high-risk retiree customer segments is a crucial issue in credit risk management, particularly because traditional scoring models are often black-box in nature and fail to provide transparent insight into the economic and behavioral factors that influence a borrower's repayment capacity. This research pursues a dual objective: first, to develop a high-performing predictive model for financial resilience classification using ensemble Machine Learning methods, and second, to apply SHAP-based Explainable AI (XAI) techniques to identify and quantify the key determinants of resilience in an accountable manner. The research methodology involved processing data from the Kaggle platform, including stratified sampling and SMOTE oversampling to address class imbalance, along with a performance comparison among Logistic Regression, Random Forest, and XGBoost. The results show that Logistic Regression proved to be the best-performing model on the held-out test set, achieving an AUC of 0.9479 and an F1-Score of 0.7313 for the resilient class. Furthermore, SHAP analysis revealed that the strongest determinants driving financial resilience were the number of prior delinquencies, loan purpose (particularly business-purpose loans), and a low debt-to-income ratio, with credit score also contributing. In practical terms, these findings provide a transparent and humane credit assessment framework, enabling financial institutions to formulate more inclusive yet prudent lending policies by prioritizing customers' actual repayment capacity over age alone.
By empirically proving that high-performance algorithms can be mathematically blind to demographic biases, this framework directly advances SDG 10 (Reduced Inequalities) and provides the accountable, feature-level justifications required for secure and sustainable financial inclusion (SDG 8).
Htet Nge Nge Ko, Aung Htoo Khine, Shadab Kalhoro et al.· Journal of Risk and Financia...· 0 citations
Predicting credit risk is vital for banks as it safeguards financial stability, minimizes default losses, optimizes capital, and ensures regulatory compliance. This study aims to predict credit risk (High/Low) in commercial banks by integrating machine learning with traditional econometric approaches. The Structural Learning in Vague Environments (SLAVE) fuzzy rule-based model handles ambiguity in financial decisions, while the eXtreme Gradient Boosting (XGBoost) uncovers non-linear patterns among predictors. Input variables—profitability, liquidity risk, ESG (environmental, social, and governance) score, and monetary freedom—were selected via multicollinearity tests and three panel regression models, including ordinary least squares (OLS), fixed effects, and random effects models. The empirical investigation uses a panel dataset of forty commercial banks across seven Middle Eastern countries from 2014 to 2023, yielding 400 observations. Regression results reveal that profitability and ESG score significantly reduce credit risk. Liquidity risk and monetary freedom increase credit risk. XGBoost combined with the SHapley Additive exPlanations (SHAP)-based interpretation identifies ESG Score as the most influential predictor. The SLAVE model was evaluated using three data splits: 70/30, 80/20, and 90/10. The 80/20 split achieved the highest accuracy, with superior performance in identifying low-risk banks. Stronger ESG performance and stable monetary environments contribute to fostering sustainable banking and reducing credit risk, making these indicators valuable for risk management frameworks in the Middle Eastern banking sector.
Jamil J. Jaber, A. A. Alkhawaldeh, Qusay Ayman Sulayman Mazahreh et al.· Risks· 0 citations
This study classifies the monthly interest rate decisions of the Central Bank of the Republic of Turkey between 2010 and 2024 into three categories: increase (hike), decrease (cut), and hold, using machine learning methods, and examines which macroeconomic and financial variables are associated with each decision category in a non-causal, prediction-focused manner. Data are obtained from the Federal Reserve Economic Data System and the electronic data delivery system of the Central Bank of the Republic of Turkey. The performance of the XGBoost, LightGBM, and CatBoost algorithms is compared against the Logistic Regression and Naive Bayes baseline models using a walk-forward validation design that preserves temporal order, with all explanatory variables lagged one month relative to the decision date to reduce contemporaneous information leakage. Class imbalance is addressed through class weighting and an adaptive oversampling procedure, and the contribution of each variable is examined using Shapley value-based explanation analysis. None of the three gradient boosting models achieves a Cohen's Kappa value meaningfully above zero in walk-forward out-of-fold evaluation, and the interest rate increase category is not correctly identified by any model; overall performance is comparable to, and in several cases weaker than, a simple majority class baseline. These findings indicate that interest rate decisions cannot be reliably predicted based on the macroeconomic indicators examined, and they support the role of institutional and communication-related factors that are not reflected in these variables. However, these findings alone do not conclusively prove this role.
Burcu Kartal· Journal of Economic Policy R...· 0 citations
Credit risk in business-to-business (B2B) transactions can threaten financial stability if it is not managed effectively. This study evaluates the use of machine learning (ML) methods to predict credit risk in B2B transactions, using 4,828 observations from large companies in Bosnia and Herzegovina over a five-year period. Three ensemble-based ML models (Bagging Decision Tree, Random Forest, and Gradient Boosting) were compared with logistic regression. All ML models showed strong predictive performance, with Gradient Boosting performing slightly better overall. Liquidity, activity, and leverage indicators were the most important predictors across all models, while non-financial variables made only a limited contribution. The findings highlight the importance of accounting information in credit risk assessment and are relevant to IFRS 9, where probability of default is a key input to expected credit loss estimation under the general approach. These results provide a basis for further development of credit risk models for non-financial companies.
Suzi Mikulić· Ekonomska Misao i Praksa· 0 citations
This study aimed to design and evaluate a hybrid financial–behavioral model integrating technical market indicators, volatility features, and social-media sentiment to improve key risk-management metrics in Bitcoin and Ethereum markets. This quantitative, ex post facto, developmental–applied study used hourly and daily financial data for Bitcoin and Ethereum from January 2019 to December 2025, together with English-language posts from Twitter/X containing relevant cryptocurrency keywords and hashtags. Financial variables included prices, trading volume, returns, volatility measures, and more than 50 technical indicators. Textual data were processed using natural language processing procedures, and sentiment scores were extracted through a finance-specific BERT model. The proposed FinBERT–LSTM model combined financial, technical, and behavioral features. Its performance was compared with naïve persistence, ARIMA, GARCH, random forest, financial LSTM, and sentiment-only models using chronological train–validation–test partitions and walk-forward validation. Predictive accuracy, directional classification, value-at-risk calibration, maximum drawdown, expected shortfall, and risk-adjusted performance were evaluated. The hybrid model significantly outperformed all benchmark models, achieving the lowest mean absolute error, root mean squared error, and mean absolute percentage error, as well as the highest directional accuracy, F1 score, and area under the curve. Diebold–Mariano tests confirmed significantly lower forecasting errors than the financial LSTM and sentiment-only models (p < 0.001). The hybrid strategy also produced lower annualized volatility, downside deviation, maximum drawdown, value at risk, and expected shortfall, while yielding higher Sharpe, Sortino, and Calmar ratios. Kupiec and Christoffersen tests indicated adequate value-at-risk coverage and independence at the 95% and 99% confidence levels. Ablation analyses further showed that removing sentiment, technical, engagement, or volatility features significantly weakened predictive and risk-management performance. Integrating technical, temporal, and behavioral information within a hybrid deep-learning architecture improves both forecasting accuracy and the management of downside and tail risk in cryptocurrency markets.
Morteza Vahdati, M. Karimi, A. Rahimi· Business, Marketing, and Fin...· 0 citations
Non-Performing Loans (NPL) are a fundamental indicator of a financial institution's asset health, reflecting loans that fail to meet interest or principal payment obligations as agreed. A high NPL ratio negatively impacts a bank's financial performance, such as decreased profitability as measured by Return on Assets (ROA) and decreased liquidity. Bank Indonesia sets an NPL tolerance limit of 5% of total credit provided by banking financial institutions. Therefore, a predictive model is needed that can detect the possibility of customers experiencing NPLs early. This study aims to identify relevant factors in predicting NPLs and create an NPL prediction model based on these factors. The contribution of this study lies in combining the results of three feature selection techniques: Chi-Square, Mutual Information, and Random Forest feature importance, using the average score eliminated by the Recursive Feature Elimination technique. Several ensemble algorithms, namely Random Forest, XGBoost, Gradient Boosting, and LightGBM, were explored to produce the best-performing model. Then, hyperparameter tuning was performed on the best model. The Random Forest model produced the best performance, with 92.17% accuracy, 78.1% precision, 98.1% recall, and 95.5% AUC. Hyperparameter tuning was shown to improve recall, thus improving the model's ability to measure how much positive data (Current class) was successfully predicted by the model. The results of this study can assist management in making credit decisions. Thus, it is hoped that it can help reduce the number of NPL cases.
David Jefri Aruan, Rusdah Rusdah, Ahmad Pudoli· IDEALIS : InDonEsiA journaL...· 0 citations
