Alternative diagnosis of corporate bankruptcy: A neuro fuzzy approach

Bankruptcy filings are as high today as ever, calling into question the efficacy of existing bankruptcy prediction models. This paper tries to provide an alternative for bankruptcy prediction by using neuro fuzzy, a hybrid approach combining the functionality of fuzzy logic and the learning ability of neural networks. The empirical results show that neuro fuzzy demonstrates a better accuracy rate, lower misclassification cost and higher detecting power than does logit regression, meaning neuro fuzzy could be a great help in providing warnings of impending bankruptcy. Also, its comprehensive explanation about mapping functions among variables presumably provides a foundation for further development of theory and testing of the membership function shape, the transfer function, the methods to aggregate, the methods to defuzzify, and so on.