A Hybrid Approach for Identification of Deficiencies in Enterprise Internal Control

Enterprise management and internal control are used to prevent and control risks and promote enterprises to achieve development strategy. This paper adopts comprehensive multidisciplinary research method to study the basic theory of prediction of major defects in enterprise internal control. Firstly, this paper proposes the prediction index system and sample selection standard of internal control major defects. Totally, 630 listed company and 12 indicators are collected and then use the random forest classification method based on principal component analysis. The parameters of random forest are optimized by genetic algorithm. Finally, the prediction model of major internal control defects of listed companies is established. The experimental results show that the average score of PCA-RF model in TPR value reaches 85%, which is nearly 20% higher than the 65% of RF model, proving that the PCA and GA can significantly improve the classification accuracy of ST Company and has important practical significance. Therefore, the proposed method system can reasonably solve the prediction problem of major defects in internal control.