• ISSN: 2010-023X (Print)
    • Abbreviated Title: Int. J. Trade, Economics and Financ.
    • Frequency: Quaterly
    • DOI: 10.18178/IJTEF
    • Editor-in-Chief: Prof.Tung-Zong (Donald) Chang
    • Managing Editor: Ms. Inez. Chan

    • Abstracting/ Indexing:  Crossref, CNKI, EBSCO

    • Article Processing Charge (APC): 500 USD

    • E-mail: ijtef.editorial.office@gmail.com

HOME > Archive > 2024 > Volume 15 Number 3 (2024) >
IJTEF 2024 Vol.15(3): 93-101
DOI: 10.18178/ijtef.2024.15.3.776

Predicting Thai Listed Company Financial Distress by Machine Learning and Synthetic Minority Oversampling Technique and Hybrid Resampling Techniques

Phatchara Plypichit 1,2 and Supranee Lisawadi 1
1. Department of Mathematics and Statistics, Thammasat University, Pathum Thani, Thailand
2. Securities and Exchange Commission, Thailand
Email: phatchara.ply@dome.tu.ac.th (P.P.); supranee@mathstat.sci.tu.ac.th (S.L.)
*Corresponding author

Manuscript received November 21, 2023; revised January 15, 2024; accepted February 7, 2024; published August 5, 2024.

Abstract—Nowadays, the techniques of machine learning have been widely adopted internationally for corporate financial distress prediction. The problem of unbalanced class distribution in classifying financial distress of listed companies on the Stock Exchange of Thailand (SET) may be addressed by implementing effective instruments by oversampling techniques and a mix of resampling methods. Many research studies have used different methods and financial ratios to classify financial distressed companies and understand the negative impact of major indicators on company financial position. This research analyzes data from financial statements gathered from 650 publicly listed firms on the SET during 2022, identify distressed companies by three consecutive years of negative earnings. Resampling was done to improve the g-mean and balanced accuracy of the three classifiers (C5.0, PART, and Generalized Linear Model (GLM)). Borderline Synthetic Minority Oversampling Technique (BLSMOTE) combined with C5.0 produces the highest median g-mean and balanced accuracy scores of 70.27% and 73.79%, respectively. In addition, SMOTE combined with Enhanced Nearest Neighbor (ENN), One-Sided Selection (OSS), and Tomek links increase the C5.0 g-mean scores compared to standalone SMOTE. Thiscomparative analysis emphasizes oversampling and hybridresampling effectiveness in addressing class imbalance infinancial distress prediction. These findings have practicalimplications for stakeholders and decision-makers, suggestingthe use of machine learning models with resampling techniquesfor early financial distress detection.


Keywords—imbalanced data, financial distress prediction, classification, resampling methods, machine learning


Cite: Phatchara Plypichit and Supranee Lisawadi, "Predicting Thai Listed Company Financial Distress by Machine Learning and Synthetic Minority Oversampling Technique and Hybrid Resampling Techniques," International Journal of Trade, Economics and Finance, vol.15, no.3, pp. 93-101, 2024.


Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).







PREVIOUS PAPER
Study on the Three Foreign Economic and Trade Wars Launched by the America after World War II
NEXT PAPER
Last page

Copyright © 2008-2024. International Journal of Trade, Economics and Finance. All rights reserved.
E-mail: ijtef.editorial.office@gmail.com