| Summary: | 碩士 === 東吳大學 === 企業管理學系 === 100 === Small and Medium Enterprise (SME) plays an important role in the process of economic development in Taiwan, and it is also the core part of business for banking. Therefore, banks are required efficient processes to prevent credit risk and to reduce NPL ratio from external and internal conditions. In the purpose, it is expect to provide an effective assessment method to evaluate risk credit level, which can decrease false rate for corporate appraisal and make appropriate decision for lending strategy.
In this paper, the focus is on the application of Random Forests (RF) to calculate powerful explanatory of financial variables and non-financial variables, applying for the main factors impact of SME credit. According to apply Rough Set Theory (RST)、Classification And Regression Tree (CART) and Decision Tree, these SME risk assessment models are established and analysis effectiveness, in order to set the best process. As the result, it is showed that using Random Forests (RF) filters and assorts the variables analyzing, enable the Risk Assessment Model is more efficient. In addition, integrated risk model can certainly reduce misjudgment of risk evaluating in SME.
|
|---|
