Credit Risk Assessment Using a Nearest-Point-Algorithm-based SVM with Design of Experiment for Parameter Selection
Credit risk assessment has become an increasingly important area for financial institutions, especially for banks and credit card companies. In the history of financial institutions, some biggest failures were related to credit risk, such as the 1974 fail
- PDF / 574,657 Bytes
- 13 Pages / 595.276 x 841.89 pts (A4) Page_size
- 25 Downloads / 186 Views
2.1 Introduction Credit risk assessment has become an increasingly important area for financial institutions, especially for banks and credit card companies. In the history of financial institutions, some biggest failures were related to credit risk, such as the 1974 failure of Herstatt Bank (Philippe, 2003). In recent years, many financial institutions suffered a great loss from a steady increase of defaults and bad loans from their counterparties. So, for the credit-granting institution, the ability to accurately discriminate the good counterparties and the bad ones has become crucial. In the credit industries, the quantitative credit scoring model has been developed for this task in many years, whose main idea is to classify the credit applicants to be good or bad according to their characters (age, income, job status, etc.) by the model built on the massive information on previous applicants’ characters and their subsequent performance. With the expansion of financial institutions’ loan portfolios, the potential benefits from any improvement of accuracy of credit evaluation practice are huge. Even a fraction of a percent increase in credit scoring accuracy is a significant accomplishment (David, 2000). So far, a great number of classification techniques have been used to develop credit risk evaluation models. They are reviewed in the first chapter. Although so many techniques were listed above, it is just a part of all for the credit risk assessment model. Some surveys on credit risk modeling gave more details about some of these techniques, such as Hand and Henley (1997), Thomas (2002), Rosenneberg and Gleit (1994), and Thomas et al. (2005). Among the different evaluation methods, the SVM approach first proposed by Vapnik (1995, 1998) achieved much good performance relative to other classification techniques. The main idea of SVM is to minimize the upper bound of the generalization error rather than empirical error.
28
2 Credit Risk Assessment Using a NPA-based SVM with DOE
Usually, SVM maps the input vectors into a high-dimensional feature space through some nonlinear mapping function. In the high-dimensional space, an optimal separating hyperplane which is one that separates the data with a maximal margin is constructed. SVM is a powerful method for classification since it has outperformed most other methods in a wide variety of applications, such as text categorization and face or fingerprint identification. After the invention of SVM, some researchers have introduced SVM to credit risk evaluation problems (Van Gestel et al., 2003; Baesens et al., 2003; Schebesch and Stecking, 2005; Wang et al., 2005; Lai et al., 2006a, 2006c). Typically, Van Gestel et al. (2003) used least squares SVM (LSSVM) (Suykens et al., 2002) for credit rating of banks and report the results contrasted with some classical techniques. Schebesch and Stecking (2005) used the standard SVM proposed by Vapnik with linear and RBF kernel for consumer credit scoring and at the same time, they used linear SVM to divide a set of labeled cre
Data Loading...
