財務危機的發生,使得企業獲利衰退,且重大財務危機的出現,更加重影響台灣上市櫃公司,近年來研究財務危機的學者,未將研究樣本依重大財務危機事件做劃分,因此本文將樣本期間區分為金融海嘯期間與金融海嘯後期間,以不同樣本期間做劃分,並以台灣上市櫃公司之一般產業為例,首先選取26個變數,利用Logistic迴歸與類神經網路在金融海嘯期間與金融海嘯後,來判別公司是否具有財務危機。其次將原始26個變數利用支援向量機進行特徵篩選,分別再使用Logistic迴歸與類神經網路,比較特徵選取後各期間樣本判別之準確率。 實證結果顯示本研究的分析方法適用於金融海嘯時與金融海嘯後,且不論樣本期間為何,類神經網路判別皆優於Logistic迴歸方法。在特徵選取上採用支援向量機是較佳的,且在區分樣本上,樣本期間越接近危機時點與增加樣本期間,都會提升財務危機預測之準確率。
This paper uses logistic regression and neural network to examine whether the companies had the financial crises during and after the periods of Financial Tsunami by collecting data from the listed companies in Taiwan. We choose 26 independent variables initially and then use recursive feature elimination based on support vector machine to select the important variables from the original 26 variables. After selecting the important variables, we use logistic regression and neural network again to determine the correct rates of prediction. Our empirical results indicate neural network methods are better than logistic regression, no matter which the sample period is. The recursive feature elimination based on support vector machine is a good feature selection. The correct rates increase, when the sample periods close to the timing of financial crises or the sample periods lengthen.