| Summary: | 碩士 === 國立臺北商業大學 === 財務金融系研究所 === 107 === The samples of this thesis are base on 142 companies suffering financial crisis and 142 normal companies with similar asset sizes were selected to be paired during the periods from 2000 to 2016. This thesis utilizes the methods of the decision tree, random forest, and logistic regression to predict the possibility of companies suffering financial crisis, and analyze to find out the financial fators causing the crisis.Our empirical results show that seven financial ratios including times interest earned, debt ratio, after-tax net profit growth rate, return on total assets, revenue growth rate, cash flow ratio and operating expense rate are able to forecast the occurrence of company’s financial distress significantly.
The prediction of the method of random forest is better than those of decision tree and logistic regression. Our empirical studies demonstrate that the F1-Score in random forest is 83.04%, the one in the decision tree is 79.99% and the one in the logistic regression is only 66.66%. In summary, the prediction of the method of random forest also performs well in recall rate, accuracy rate and precision rate.
|
|---|
