Construction of Earnings Management Prediction Models

• Main Authors: WANG, HAN-SHENG, 王瀚陞
• Other Authors: CHI, DER-JANG
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/c6phf9

碩士 === 中國文化大學 === 會計學系 === 106 === The purpose of this study is to establish an earnings management prediction model for enterprises. The data is from the Taiwan Economic Journal (TEJ), from 2009 to 2016 of the biotechnology industry. Most of the previous literatures use traditional regression models on earnings management. In recent years, many researchers try to use data mining methods to improve the accuracy of earnings management prediction. Therefore, this study uses data mining to construct prediction models. In the first stage of this study, support vector machine (SVM), and artificial neural network (ANN) are used to select important variables. In the second stage, decision trees-CART, CHAID and CHAID are used to establish models. The empirical results of this study show that the ANN-C5.0 model has the highest prediction accuracy of 88.03%.