應用資料探勘於盈餘管理預測

CCUR > College of Business > Department of Accounting & Graduate Institute of Accounting > Thesis > Item 987654321/33701

Please use this identifier to cite or link to this item: https://irlib.pccu.edu.tw/handle/987654321/33701

Title:	應用資料探勘於盈餘管理預測 An Application of Data Mining Techniques to Predict Earnings Management
Authors:	高毅
Contributors:	會計學系
Keywords:	應計項目盈餘管理資料探勘多層感知器決策樹支援向量機 discretional accruals earnings management data mining multilayer perceptron decision tree support vector machine
Date:	2016-06
Issue Date:	2016-08-17 14:02:39 (UTC+8)
Abstract:	企業所發布之財務報表上的攸關性及可靠性與管理階層操縱裁決性應計項目的盈餘管理行為多年一直是重大議題。過往文獻於裁決性應計項目盈餘管理上皆是使用傳統的回歸模式，近年來已有許多學者應用資料探勘的方法針對應計項目盈餘管理進行研究，然而準確度皆有所提高，因此本研究以傳統逐步回歸與資料探勘法中的支援向量機以及類神經網路來進行預測，希望能比較出傳統與資料探勘準確度差異，以及找出一個較準確地預測模式與規則。本研究嘗試以支援向量機(support vector machine)、多層感知器(multi-layer perceptron, MLP)以及逐步回歸(stepwise regression , STW)先將變數進行第一階段的篩選，再進一步使用決策樹CHAID及決策樹C5.0來建立模型檢測企業是否具有嚴重操縱盈餘的情況。而實證結果顯示，支援向量機搭配決策樹CHAID表現最佳，準確率為97.63%。 Companies release and reliability financial statements and management manipulated discretional accruals earnings management on the issue has been a major accounting. Past literature on the discretional accruals earnings management are all using traditional re- gression model, in recent years there have been many scholars have applied data mining methods for discretional earnings management research. Therefore, this study traditional stepwise regression and data mining methods neural network to predict, hoping to com- pare the accuracy of traditional data mining differences, and to identify a more accurate prediction models and rules. This study attempts to regression (stepwise regression,STW)、(support vector machine, SVM) and multilayer perceptron (multi-layer perceptron, MLP) first stage of the first variable filter, further use of decision trees CHAID and deci- sion tree C5.0 to model detection whether a company has serious manipulate earnings. The decision tree CHAID demonstrates the best performance for earnings management with the accuracy rate of 97.63%.
Appears in Collections:	[Department of Accounting & Graduate Institute of Accounting] Thesis

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	308	View/Open

View Licence

Loading...