Predicting Risk Parameters Using Intelligent Fuzzy/Logistic Regression

A new hybrid approach based on fuzzy logics and logistic regression analysis is proposed to demonstrate whether supervision has early warning ability of financial crises for Taiwan´s banking industry. First, it extracts significant financial variables, macroeconomic variable and corporate governance variable affecting the occurrence of banking crises from past literature, and then creates financial crisis warning system with data of 5 years, 4 years, 3 years, and 2 years before the occurrence of banking crises and discussed key factors causing banking crises. Extracted key factors are further predicted by a fuzzy logics model. When the fuzzy forecasting model is developed to predict the key variables, satisfactory results are obtained. The current hybrid approach is proved to be robust and accurate.