| 摘要: | 本研究探討財務危機預警模型的構建是一個受到廣泛關注之研究課題，其所帶來的影響不僅是發生在公司內部亦會影響到公司外部的所有人，如何構建一個有效的預警模型在會計、財金與管理學門上皆為相當重要的任務，大部分財務危機預警模型的構建皆採用單一之型態，但任何之預警模型皆有其先天的限制，為了克服單一模型之限制，本研究基於整體學習之概念發展出一個新穎的整合模型，其包含三大構面：(1)資料前處理(T檢定)、(2)模型效力的提升(K-means)、(3)最終模型的構建(SVM)。為了進一步驗證本預警模型的效力，將其應用於台灣財務危機公司之預警，研究結果指出本模型具有相當優越的泛化能力與預測效果，此預警模型可以提供給決策者做有效之判斷。
Financial crisis forecasting is an essential and widely researches domains since it can have considerable effect on inner and outside parts of companies. How to effectively predict financial crisis is an important task in accounting, finance and management. Though much attention has been paid to financial crisis forecasting approaches based on singular classifier, its limitation of uncertainty and advantage of hybrid mechanism for financial crisis forecasting has also been ignored. Inspired by ensemble learning, the study introduced an emerging hybrid mechanism which hybrid t-test, K-means and Support Vector Machines (SVM) for financial crisis forecasting. According to our empirical results, the proposed hybrid mechanism poses outstanding performance in terms of forecasting accuracy. The proposed mechanism can give a proper direction for decision makers to make a judgment in turbulent economic environment. |
