Predicting financial failure in corporations listed on the Amman Stock Exchange using machine learning

This study aims to investigate the effectiveness of integrating Machine learning (ML) algorithms with the Altman model and to evaluate their performance in predicting financial failure. Moreover, the performance of the proposed ML model will be compared with that of the traditional Altman Z-score model. The study population consists of Jordanian public shareholding companies in the industrial and service sectors. The period of this study from 1999 to 2024. Several American companies that declared bankruptcy in 2024 considered to increase the population size to achieve reliable accuracy. This sample is analyzed using the Altman Z-score model. Subsequently, historical financial data used in the Altman model equations, without applying the weights, were used to train the machine learning model (using the H2O AutoML platform). The performance of both models then measured using the confusion matrix, recall, accuracy, precision, and the F-measure. The results showed that the Altman model achieved 77.61% accuracy, while the developed machine learning model (using H2O AutoML) achieved 89.47% accuracy, using ensemble machine learning (XRT-1, DRF-1, and a stacked ensemble model). The study found that the developed machine learning model outperforms the Altman model across accuracy, recall, precision, and F-measure.