Document is current

Article History

Received: 1 May 2026

Accepted: 12 August 2026

First Online: 13 August 2026

Declarations

Competing interests: The authors declare no competing interests.

Ethical approval: This article does not report research involving human participants or animals. The study complies with international ethical standards for academic publishing.

Reproducibility statement: To improve reproducibility, the revised manuscript reports the preprocessing pipeline, chronological data split, feature-partitioning strategy, model architecture, hyperparameter search space, selected hyperparameters, training configuration, evaluation metrics, and explainability procedure in detail. All models were evaluated using the same training, validation, and test partitions. The test set was not used to estimate preprocessing parameters, optimize hyperparameters, select models, or tune the decision threshold. The implementation was developed in Python using TensorFlow/Keras, scikit-learn, Optuna, SHAP, and LIME.