Corporate credit scoring is important for investors and banks in risk management. However, the high dimensional data available from public financial statements make credit analysis difficult. To address the problem, dimensionality reduction is a key step to enhance scoring accuracy. By using semi-supervised discriminant analysis (SSDA) and support vector machines (SVMs), this study develops a novel system for credit scoring, where SSDA transforms high dimensional data space (over 50 financial variables) to a perfect low dimensional representative subspace with maximal discriminating power. Constructing SVM classifier in the new space effectively reduces overfitting and enhances classification accuracy. Empirical results indicate that SSDA is better than traditional dimensionality reduction schemes, and it significant improves SVM performance. More importantly, the new classification system substantially outperforms conventional classifiers. The new decision support system can help corporate bond investors make good assessments on their risks and substantially reduce their losses.
Key words: Semi-supervised discriminant analysis, dimensionality reduction, credit scoring, support vector machine, risk management.
Copyright © 2022 Author(s) retain the copyright of this article.
This article is published under the terms of the Creative Commons Attribution License 4.0