데이터마이닝 기법을 적용한 내부회계취약기업 특성 식별 및 예측에 관한 연구

This study tested the applicability of a data mining tool in the internal control deficiencies discrimination and to develop internal control deficiencies prediction models. I am applying data mining techniques including decision tree(DT), neural network(NN), ensemble techniques. This study uses the financial measures that are size, ROA, inventory・receivables/asset, loss dummy, debt, credit rating score, total accruals and nonfinancial measures are that are volatility, beta, owner, ownership percentage of the largest shareholder, ratio of the number of employees involved with the implementation of internal controls, firm’s listed age, non-audit fee. This study analysed KOSPI and KOSDAQ firms for the period from 2005 to 2013. Through logistic regression analysis, ROA, loss dummy, ownership percentage of the largest shareholder are included in the prediction model. And Through decision tree analysis, ROA, inventory・receivables/asset, debt, credit rating score are included in the prediction model. Based on small mis-classfication ratio, ensemble techniques are higher than those of logistic regression, decision tree, and neural network . In conclusion, this study reports the positive results of an analyses examining the effects of internal control deficiencies discrimination and prediction on the effectiveness of data mining. Especially, ensemble techniques are better then single technique. And these results have implications for the internal control assess of stakeholder to do the tasks and the structuring of these tasks.