Research on Financial Crisis Forewarning Model Based on Rough Set Theory and Bayesian Neural Network

Multivariate discriminant analysis for (MDA), Logistic regression analysis and other traditional methods for financial crisis forewarning need to satisfy statistical hypothesis. BP neural network tends to form a local minimum, and has problems of slow convergence and over-training. Then, this paper establishes a practical financial crisis forewarning model based on rough set theory and Bayesian Slovakia neural network. By collecting financial data of listed companies, this paper uses SPSS software to complete data preprocessing, uses excel software to discretize continuous data, uses Rosetta software to reduce property, and then sets up the Bayesian regularization neural network model. Predicted results show that the training speed and error rate are satisfying.