Early Warning and Prevention of non-Compliance of Internal Control Information Disclosure based on data Mining

High quality internal control information disclosure can promote the healthy development of the capital market. This paper selects the Shanghai and Shenzhen A-share listed companies in 2017 as the research object, and selects the data of corporate finance, corporate governance, investor protection, market and executive characteristics and incentives to construct an early warning index system with the combination of financial and non-financial indicators, and uses Bayesian classification, Logistic regression, decision tree and K-proximity learning to predict the non-compliance behavior of internal control information disclosure. The study finds that after feature attribute selection, the prediction accuracy of non-compliance behavior is significantly improved, the highest is logistic regression model; secondly, it is found that non-financial information contributes more than financial information in the prediction process. Therefore, the integrity and self-discipline management of the main body of the enterprise, the external maintenance of high standards of professional ethics, and all-round efforts can jointly build and share a high-quality benign capital market environment.