Research On The Prediction Model Of Material Weaknesses In Internal Control Of Listed Companies Based On Support Vector Machine

The disclosure proportion of material weaknesses in internal control,listed companies in China is far lower than the United States.The reason may be that,because of our country's imperfect identification standards,listed companies or auditors does not recognize material weaknesses,or the management found the material weaknesses but not choose to tell the public.Based on the disclosure of financial data and regulatory information,prediction model of material weakness in internal control use some prediction technique to identify and predict the possibility of existing material weaknesses.Before China's internal control system been perfect and effective and after a period of time,to predict the possibility of existing material weaknesses will help the company improve the internal control environment to avoid great losses.According to the predict results,investors can assess the company's risk and make the right investment decisions,supervision departments can pay attention to the company which may have material weaknesses to save the cost of supervision and improve the supervision efficiency.It has important practical significance.This paper builds the prediction index system of material weaknesses in internal control that including four parts:the complexity of operation,organizational structure change,accounting risk and financial risk,and internal control construction.Based on China's related data in 2012 to 2015,we selected 10 effective prediction indexes.Then according to the selecting standards which we made,we selected the material weaknesses samples and control samples and divided into training samples and test samples.through the use of support vector machine to training and testing,we got the final test results:the overall recognition rate is 82.67%,the control samples recognition rate is of 83.82%,and the material weaknesses samples recognition rate is 52.81%.The results show that:the prediction index system of material weaknesses in internal control including the number of industries involved in the business,if-is-loss,Z value,the total non recurring gains and losses,rate of return on net assets,operating income growth rate,listed years,if-is-state-owned,the proportion of top ten shareholders and if-audit-institutions-changes.This paper provides a support vector machine prediction method which based on the prediction index system of material weaknesses in internal control.It can use the financial data and monitoring information to identify and predict the material weaknesses in internal control.