Satellite-Inspired Time-Frequency Deep Learning for Predicting and Assessing Financial Health in General Contractors

The ability to accurately predict and evaluate financial health is crucial for project owners and stakeholders to ensure stability and minimize risk. Traditional models such as the Altman Z-score, while widely used, adapt poorly to rapidly changing economic conditions. To address this limitation, this study develops the novel hybrid deep learning financial health assessment model for general contractors (FHAM-GC), which integrates the Altman Z-score, grey relational analysis, and artificial satellite search algorithm–revolving gate Fourier transform to capture complex time-series dynamics in financial data. The model was trained and validated on 12 years of quarterly financial statements from general contractors listed on the Indonesia Stock Exchange (IDX). The comparative tests conducted on the proposed model result in a reference index of 0.9921 and an R2 of 0.9738 in the testing phase, with a mean absolute percentage error of only 4.67%. These results indicate the proposed model outperforms optimized hybrid, hybrid, and basic deep learning models by approximately 2%, 10%, and 40%, respectively, confirming the superior predictive capability and robustness of the proposed model across all evaluation metrics. A notable strength of the proposed model is its ability to maintain high accuracy and stability across time-series forecasting tasks, ensuring the regular generation of reliable financial health predictions. This comprehensive framework provides stakeholders of IDX-listed Indonesian general contractors, including owners, management teams, shareholders, and clients, with a reliable financial health monitoring model that enables the proactive tracking of financial performance trends and enhances overall decision-making in the construction industry. The FHAM-GC developed in this study supports continuous financial health assessment to facilitate sustainable industry stability and informed strategic planning.