Deveopment of Optimal Financial Distress Prediction Model based on Data Mining Techniques
| Autor: | Wei-ming Chiu, 邱偉明 |
|---|---|
| Rok vydání: | 2007 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 95 Due to recent informational globalization, it unleashes a major change of how the business should operate under today’s environment. Meanwhile, the worsening macroeconomic conditions have increased the possibility of financial distress. To investors of businesses, the main reason to invest additional capital into the market is determined by whether or not the business can continue to survive in the current market. At the same time, the financial standing of a business can be the main reason for the given security of a financial institute to loan the necessary capital for investment. To financial institutes, predicting financial distress accurately can not only further increase the efficiency of capital distribution, but also decrease the possibility of bad debts. The aim of this thesis is to apply machine learning techniques to develop an optimal hybrid model. To examine it applicability, the hybrid model is compared with different single and multiple classifiers in terms of prediction accuracy and related error rates. By using five bankruptcy prediction and credit scoring datasets, the hybrid model based on combining Self-Organizing Maps (SOM) with multiple classifiers performs the best, especially for multiple homogeneous classifiers. The hybrid model provides higher predication accuracy and lower Type I & II errors. In addition, some fuzzy data, which cannot be recognized by SOM are used to test these models. The hybrid model still performs the best. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|