Corporate Fraud Prediction：The Application of Data Mining Techniques

• Main Authors: yu-chi chang, 張育綺
• Other Authors: Yu-Hsin Lu
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/96361067303682634405

碩士 === 逢甲大學 === 會計學系 === 102 === The purpose of this study is to employ the feature selection by using Association Rules (AR) a data mining technique and to choose the important affecting factors of corporate fraud prediction. The empirical result indicates that the prediction performance of feature selection is better than without feature selection. The critical factors extracted out are characterized by the scale of directors and supervisors, independent directors and supervisors seats, major shareholders and used in Decision Tree (DT) classification technology to construct prediction model. The empirical results provide that overall prediction accuracy was 79.17%, and indicate the decision table of corporate fraud prediction. The study expect these results can provide the desired decision table to the user, and reduce losses from the investments in future.