VGU RESEARCH REPOSITORY
Please use this identifier to cite or link to this item:
https://epub.vgu.edu.vn/handle/dlibvgu/2008| Title: | Comprehensive Machine Learning Applications in Corporate Bankruptcy Risk Prediction: Evidence from Vietnam | Authors: | Nguyen Dinh Anh Khoa | Keywords: | bankruptcy risks, machine learning, Elastic Net, LASSO, neural network | Issue Date: | Dec-2024 | Source: | IEEE | Abstract: | This study aims to comprehensively examine bankruptcy risk prediction in Vietnam by applying and evaluating the performance of several machine learning models, such as neural networks and tree-based algorithms, alongside advanced feature selection techniques. This is the first study to apply Elastic Net for feature selection in this context, alongside LASSO regression. Utilizing financial data from over 1500 publicly traded companies on Vietnam stock exchanges, the largest dataset in the field to date, the study explores the applicability of predictive modeling in predicting corporate bankruptcy risks and identifying key financial indicators of financial distress. The findings reveal that most models performed very well despite the class imbalance and missing values within the dataset. The study also demonstrates that Elastic Net outperform LASSO in feature selection by retaining more relevant features and maintaining models’ stability, particularly with kNN, SVM, and LSTM. Key financial metrics, including debt ratios, profitability measures, liquidity indicators, and market confidence variables, emerge as significant predictors. Moreover, findings also suggest missing values may carry valuable insights in predicting financial distress. These obtained results have practical implications for credit risk assessment, early warning systems for regulators, and corporate financial planning. |
URI(1): | https://epub.vgu.edu.vn/handle/dlibvgu/2008 | Rights: | Attribution-NonCommercial 4.0 International |
| Appears in Collections: | Finance & Accounting (FA) |
Files in This Item:
| File | Description | Size | Format | Existing users please Login |
|---|---|---|---|---|
| Thesis v3.pdf | 1.69 MB | Adobe PDF |
Page view(s)
221
checked on Nov 15, 2025
Download(s)
26
checked on Nov 15, 2025
Google ScholarTM
Check
This item is licensed under a Creative Commons License