| Summary: | 碩士 === 國防大學管理學院 === 財務管理學系 === 100 === Financial crises often pay high social costs. The alarms and the causes of the crisis is also the recently continuing concern issue. This study attempts to build up a framework of financial crises alarm by extracting key financial ratios and alarm rules applied data mining techniques and factor analysis. Through comparison of the classification error rate among the different decision tree models, this paper summarizes the results to integrate an optimal predictive model. According to the model, the financial crisis firms will be discovered in advance by screened to the best prediction model. To validate the racticality of the financial distress prediction model, this paper referred to the operating rules of the Taiwan Stock Exchange Corporation (TSEC) and collected the listed companies as the initial sample from 2001 to 2009. The study adopted the pairwise sample of crisis and normal firms selected for training and testing data, and selected the previous fourth-season data before crises occurred. The study found the prediction model established by screening financial ratios can effectively reduce the type II error rates, which also improve the overall prediction accuracy rate. The model and results as the reference of investment, the Public Employees Pension Fund could increase revenues and avoid the risk resulting from the wrong investment.
|
|---|
