A neural network for classifying the financial health of a firm

We present here a neural network applied to a universal business problem: the estimation of the future fiscal health of a corporation. The commonly used accounting and financial tool for such classification and prediction is a multiple discriminant analysis (MDA) of financial ratios. But the MDA technique has limitations based on its assumptions of linear separability, multivariate normality, and independence of the predictive variables. A neural network, being free from such constraining assumptions, is able to achieve superior results. Our neural network model is the Cascade-Correlation architecture recently developed by Scott E. Fahlman and Christian Lebiere at Carnegie Mellon University. This new approach solves the hidden architecture enigma encountered using other types of neural networks. Also, Cascade-Correlation manages error signals in a manner which significantly improves execution speed. Our research is the first to use Cascade-Correlation for corporation health estimation.