地方财政收入的组合预测模型研究

由于地方财政收入预测受到诸多因素的影响,而现有的预测方法不尽科学,因此提出了二次组合预测方法。首先通过灰色关联分析确定影响较大的外生变量,然后利用神经网络对选择出的外生变量的预测结果进行组合预测,最后将该组合预测结果与单项预测方法加权平均形成二次组合预测模型。实验证明该方法在地方财政收入预测中是切实可行的。

Research on Combination Forecasting Model of the Local Financial Revenue

The local financial revenue is affected by many factors and there are some disadvantages in current forecasting methods.Therefore,the secondary combination forecasting model is put forward.First,the main factors for local financial revenue are confirmed via gray correlation analysis.Then the combination forecasting results from neural network and the first step＇s re-sults is obtained.In the end,by weighed average function,the first combination forecasting result is combined with the re-sults of the methods using only one algorithm,and the secondary combination forecasting model is created.Experimental re-sults demonstrate the availability and feasibility of the model in the local financial revenue forecasting.