應用資料探勘技術建構繼續經營預測模型

CCUR > College of Business > Department of Accounting & Graduate Institute of Accounting > Thesis > Item 987654321/44303

Please use this identifier to cite or link to this item: https://irlib.pccu.edu.tw/handle/987654321/44303

Title:	應用資料探勘技術建構繼續經營預測模型 Construction of Going Concern Prediction Models by Using Data Mining Approach
Authors:	吳羿君 (WU, YI-JUN)
Contributors:	會計學系
Keywords:	繼續經營資料探勘類神經網路支援向量機決策樹 going concern data mining artificial neural network(ANN) support vector machine(SVM) decision tree
Date:	2018
Issue Date:	2019-05-29 12:58:13 (UTC+8)
Abstract:	投資大眾及利害關係人若未即時發現企業破產發生之可能性，將可能造成在做投資行為或重要決策時判斷錯誤，導致重大損失或無法挽回之結果。近年來有許多學者使用資料探勘法進行繼續經營決策意見相關研究，都有相當不錯的準確率，但整體文獻還不夠完整。而本研究的目的在建立一個嚴謹的繼續經營預測模型，研究資料來自台灣經濟新報，研究對象為2009年至2016年間，有被出具繼續經營疑慮意見及未被出具繼續經營疑慮意見之上市上櫃公司，以一比三進行配對。本研究第一階段以資料探勘中的類神經網路以及支援向量機來篩選出重要變數，配合C5.0、CHAID以及QUEST三種決策樹演算法分別建構分類模型並進行比較，變數方面則採用財務及非財務變數，以期望能得出更精準的繼續經營預測模式。研究結果發現，以支援向量機篩選重要變數，搭配 C5.0 決策樹演算法之分類效果最好，準確率達90%。 In recent years, many researchers use data mining on going concern research to improve the prediction accuracy. The purpose of this study is to establish rigorous going concern prediction models. The research data is from the Taiwan Economic Journal (TEJ) from 2009 to 2016. The listed companies with going concern doubts and without going concern doubts are paired with 1:3. In the first stage of the study, ANN, and SVM are used to filter out important variables. In the second stage, decision tree-C5.0, CHAID, and QUEST are used to build prediction models. The empirical results of this study show that the SVM-C5.0 is the best going concern prediction model with an accuracy of 90%.
Appears in Collections:	[Department of Accounting & Graduate Institute of Accounting] Thesis

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	165	View/Open

View Licence

Loading...