National Changhua University of Education Institutional Repository : Item 987654321/17072
English  |  正體中文  |  简体中文  |  全文筆數/總筆數 : 6507/11669
造訪人次 : 30010960      線上人數 : 396
RC Version 3.2 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
搜尋範圍 進階搜尋

請使用永久網址來引用或連結此文件: http://ir.ncue.edu.tw/ir/handle/987654321/17072

題名: Integrating GA with Boosting Methods for Financial Distress Predictions
整合GA與Boosting演算法於財務危機之預測
作者: Liu, Hsin-Yu;Huang, Shian-Chang
貢獻者: 企業管理學系
關鍵詞: Financial distress;Boosting algorithm;Genetic algorithm;AdaboostM1;Logitboost;Multiboost
財務危機;Boosting演算法;基因演算法;AdaboostM1;Logitboost;Multiboost
日期: 2010-04
上傳時間: 2013-07-11T09:04:31Z
出版者: 中華民國品質學會
摘要: Financial distress is the most considerable and notable distress for companies. It also has a direct effect on its development and survival, and may result in a crisis in capital markets. Thus, financial distress prediction has been a critical issue in the area of academia and industry. The aims of this study are two folds: first, to compare prediction algorithms from data mining with traditional statistical methods, and second, to combine genetic algorithms (GAs) with boosting methods for developing a reliable and accurate model of bankruptcy prediction. The base classifiers we used are decision trees, logistic regressions, neural networks, and support vector machines. The boosting algorithms used are AdaboostM1, Logitboost, and Multiboost. The above algorithms are optimized by GA for input features. Empirical results indicated that integrating GA with AdaBoostM1 achieves the best performance.
財務危機是公司最嚴重且最引人注意的風險。它直接影響公司的發展和生存，也可能導致資本市場的危機發生。因此，無論是在學術界或是實務界，財務危機的預測一直是個重要的議題。本研究主要有兩個階段：首先，比較資料探勘法和傳統統計方法。其次，使用Boosting演算法結合基因演算法建立一個可信且準確的危機預測模型。本研究以決策樹、邏輯斯迴歸、類神經網絡和支援向量機作為單一分類器。Boosting演算法則採用AdaboostM1、Logitboost和Multiboost。所有演算法藉由基因演算法來選擇適配的特徵變數。實證結果顯示AdaBoostM1結合基因演算法有最好的表現。
關聯: Journal of Quality, 17(2): 131-158
顯示於類別:[企業管理學系] 期刊論文

文件中的檔案:

檔案 大小格式瀏覽次數
index.html0KbHTML649檢視/開啟


在NCUEIR中所有的資料項目都受到原著作權保護.

 

DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - 回饋