Article: ANS-REA algorithm for dynamic corporate financial distress prediction Journal: International Journal of Reasoning-based Intelligent Systems (IJRIS) 2026 Vol.18 No.10 pp.31 - 45 Abstract: Traditional prediction models often overlook the complexity of enterprise data, resulting in unstable prediction results. This study proposes a dynamic corporate financial distress prediction model based on the adaptive neighbour synthesis minority oversampling technique-recursive integration method. The area under the receiver operating characteristic curve (AUC) is adopted as the primary metric for evaluating prediction accuracy. The sample covered 2,850 listed companies, and the data collection period was from 2014 to 2022, involving seven major industries. The results show that the classifier algorithm based on random forest achieves an accuracy of 91.38%. The proposed algorithm achieves an accuracy of 91.96% when dealing with imbalanced data, and the prediction model combined with five time periods achieves an accuracy of 92.5%. The results show that the prediction model based on the adaptive neighbour synthetic minority over-sampling technique-recursive integration approach can provide a potential tool for corporate risk assessment. Inderscience Publishers - linking academia, business and industry through research

Title: ANS-REA algorithm for dynamic corporate financial distress prediction

Authors: Bo Wang

Addresses: School of Economics, University of Chinese Academy of Social Sciences, Beijing 100836, China

Abstract: Traditional prediction models often overlook the complexity of enterprise data, resulting in unstable prediction results. This study proposes a dynamic corporate financial distress prediction model based on the adaptive neighbour synthesis minority oversampling technique-recursive integration method. The area under the receiver operating characteristic curve (AUC) is adopted as the primary metric for evaluating prediction accuracy. The sample covered 2,850 listed companies, and the data collection period was from 2014 to 2022, involving seven major industries. The results show that the classifier algorithm based on random forest achieves an accuracy of 91.38%. The proposed algorithm achieves an accuracy of 91.96% when dealing with imbalanced data, and the prediction model combined with five time periods achieves an accuracy of 92.5%. The results show that the prediction model based on the adaptive neighbour synthetic minority over-sampling technique-recursive integration approach can provide a potential tool for corporate risk assessment.

Keywords: ANS-REA; financial distress; random forest; unbalanced data; AUC.

DOI: 10.1504/IJRIS.2026.152544

International Journal of Reasoning-based Intelligent Systems, 2026 Vol.18 No.10, pp.31 - 45

Received: 04 Nov 2025
Accepted: 30 Dec 2025
Published online: 26 Mar 2026 *

Title: ANS-REA algorithm for dynamic corporate financial distress prediction

Keep up-to-date