中央预算执行审计对象遴选方法探索——基于GRNN神经网络的部门预算违规率预测

The Object Selection of Central Budget Implementation Audit——Basing on GRNN Forecast of the Departments' Budget Violations Rate

在中央省部级单位众多、审计署人力物力资源有限的情况下,探索在年度中期实现对中央预算执行审计对象当年预算违规状况的预测,并据此进行审计对象的科学遴选,这对于提高预算执行审计效率具有一定的现实意义。本文利用MATLAB软件和2009年—2014年中央各部门预算执行审计数据构建了GRNN神经网络,对2015年各部门年预算违规比率进行预测,并与真实违规率进行比较,结果表明GRNN具有较好的预算违规预测能力。

China＇s Central government has many ministerial departments, and the resources of CNAO are limited. Based on this situation, exploring the budgets illegal status prediction of the central budget implementation audit＇s object in the current year, making scientific selection of audit object according to the prediction, for upgrading the efficiency of budget implementation audit, has certain practical significance. Using MATLAB software and the central departments budget implementation audit data from 2009 to 2014 to construct the generalized regression neural network （GRNN）, and carry on the forecast of the 2015 each department＇ s budget violations rate. Comparing the rate with real violation rate, it has found that GRNN has good budget viola- tion forecast ability.