In the different phase of the business cycle, companies may face different financial crises with different financial attributes. In this paper we take the phases of the business cycle into consideration and use data mining as a technique to predict firms that may face a potential
financial crisis. Since we hypothesize that financial crises are closely related to the business cycle, we have determined which periods show expansionary or recessionary trends. The objective of this paper is to determine important variables affecting financial crises and use this information to improve the accuracy of financial predictions. The electronic industry has been selected as the main focus, as this is a very important industry in Taiwan. Five experimental models have been designed for empirical study and an optimal model has also been established. From the results, we have discovered some important financial variables that can cause business’s financial crises in different phases of the business cycle. In considering the business cycle, the model achieved better predictive accuracy and a support
vector model has a higher predictive accuracy than other data mining techniques.