Qualitative Input Conditioning to Enhance RBF Neural Networks Generalization in Financial Rating Classification


Xavier Parra, Xavier.Parra@upc.es ESAII, Universitat Politècnica de Catalunya Av. Víctor Balaguer, s/n – 08800 Vilanova i la Geltrú – Barcelona (Spain)

Núria Agell, agell@esade.edu ESADE, Universitat Ramon Llull Av. Pedralbes, 62 – 08034 Barcelona (Spain)


Abstract: The rating is a qualified assessment about the credit risk of bonds issued by a government or a company. There are specialized rating agencies, which classify firms according to their level of risk. These agencies use both quantitative and qualitative information to assign ratings to issues. The final rating is the judgement of the agency’s analysts and reflects the probability of issuer default. Since the final rating has a strong dependency on the experts knowledge, it seems reasonable the application of learning based techniques to acquire that knowledge. The learning techniques applied are neural networks and the architecture used corresponds to radial basis function neural networks. A convenient adaptation of the variables involved in the problem is strongly recommended when using learning techniques. The paper aims at conditioning the input information in order to enhance the neural network generalization by adding qualitative expert information on orders of magnitude. An example of this method applied to some industrial firms is given.

Keywords: Orders of Magnitude, Neural Networks, Credit Risk.


Full Paper (PDF 88.7 KB)