Evaluating corporate failure risk with a new intelligent processing approach

Corporate failure is an important issue to better understand, and, if possible, predict. In this paper we propose using a new intelligent processing method-the hierarchical multiple-feature fuzzy neural (HMFN) approach-to implement the modelling of failure risk evaluation. The resultant model has been evaluated by comparison with the performances of optimised conventional neural network models. It is found that, being capable of processing a wider range of information (both quantitative and qualitative) as well as of coping with subjective inference, the HMFN model fan achieve better quality in terms of accuracy, explanation capacity, and generalisation ability.