Summary: | 碩士 === 朝陽科技大學 === 保險金融管理系碩士班 === 96 === This paper adopts data mining technology to explore the factors of loan default in domestic bank. Based on the misclassification cost of Bortiz and Kennedy(1995), the performances between different models with cut-off value are compared. The results revealed that the model with 0.6 cut-off value obtained relatively minimum misclassification cost. Finally, the significant variables that affected the loan default were loan amount, interest rate, revenue growth in recent two years, and owning real estate or not. More specifically, the revenue growth in resent two years is the most significant variable.
|