Abstract:

The purpose of the study is to test the failure forecasting by using the hybrid techniques of Genetic Algorithm (GA) and Adaptive Neuro Fuzzy Inference System (ANFIS). The key factors of failure signal are 21 financial ratios which are calculated from 109 Thai enterprises. This study separates the data set into 2 categories which are 70:30 training set and test set respectively. At the beginning of the study process, GA is used to select attribute from the training set for serving ANFIS to generate the forecasting model. Another set of data is used to prove the accuracy rate and error rate of testing result. According to the result of the study, the result points out that 5 financial ratios from GA-ANFIS selection have ability to reform the failure estimation model with 90.63 percent of accuracy rate and have 3.13 percent of type 1 error rate. Therefore, this study concludes that model from GA-ANFIS selection is very useful to predict the failure of Thai enterprises affirmatively.