ROUGH SETS AND DISCRIMINANT ANALYSIS TECHNIQUES FOR BUSINESS DEFAULT FORECASTING
José David Cabedo. Department of Finance and Accounting, Universitat Jaume I
José Miguel Tirado. Department of Finance and Accounting, Universitat Jaume I
- Fuzzy Economic Review: Volume XX, Number 1. May 2015
- DOI: 10.25102/fer.2015.01.01
Abstract
One area for the application of rough sets theory is business failure prediction. Taking a set of financial ratios as the starting point, the decision rules generated from the in-the-sample set of companies can be used to forecast the default/healthy situation of the out-of-the-sample set companies. Some companies, however, cannot be allocated to the healthy or the default set. In this paper we propose the joint use of rough sets theory and discriminant analysis techniques. We use the theory to generate decision rules and we then use discriminant analysis techniques for companies that cannot be clearly allocated to a decision class. Our proposal does not require the involvement of an expert to solve this company allocation problem, thereby overcoming the drawbacks of other alternatives when they must be integrated into the organisation’s standard procedures (i.e. those involving the concession of a credit facility in a bank). We have applied our proposal to a sample of Spanish nonfinancial corporations and show how our results are an improvement on application of plain vanilla discriminant analysis.