基于SARIMA和Elman模型及其组合模型的财政支出研究

发布时间：2018-04-30 06:12

本文选题：SARIMA模型 + Elman模型　；参考：《兰州大学》2017年硕士论文

【摘要】：随着经济的发展,人民生活水平得到显著的提高,中国已成为世界上第二大经济体,国内生产总值连年递增,国家财政支出也呈逐年上升的趋势。财政支出是政府实施宏观调控,进行资源优化配置的有效手段。近年来,国家财政在民生方面的支出比例不断增加,人民切实感受到了财政支出所带来的实惠。由于财政支出数据具有明显的季节性特征,因此在进行数据建模时,要先对原始序列进行季节性调整。为了更好地预测财政支出的增长趋势,本文首先建立了季节性差分自回归滑动平均模型(SARIMA)和Elman神经网络模型,结果表明SARIMA模型要优于Elman模型。其次,又建立了 SARIMA模型和Elman模型相结合的组合模型,并对国民财政支出数据进行最终的预测。在组合模型中,为了寻找最优的组合系数,我们采用了粒子群优化算法(PSO),以平均相对百分比误差(MAPE)为目标函数,使组合模型的预测误差最小。最后,我们实证分析了 2000年1月-2015年12月财政支出序列,并对2016年的国家财政支出进行预测,结果表明组合模型的效果优于单一模型的效果,从而为我国财政支出预算提供了科学依据。
[Abstract]:With the development of economy, the living standard of the people has been greatly improved. China has become the second largest economy in the world, the GDP has been increasing year after year, and the national fiscal expenditure is also rising year by year. Fiscal expenditure is an effective means for the government to implement macro-control and optimize the allocation of resources. In recent years, the proportion of government expenditure on people's livelihood has been increasing, and the people really feel the benefits brought by fiscal expenditure. Due to the obvious seasonal characteristics of fiscal expenditure data, the original series should be seasonally adjusted when modeling the data. In order to better predict the increasing trend of fiscal expenditure, the seasonal differential autoregressive moving average model (SARIMA) and the Elman neural network model are established in this paper. The results show that the SARIMA model is superior to the Elman model. Secondly, the combination model of SARIMA model and Elman model is established, and the final prediction of national financial expenditure data is made. In order to find the optimal combination coefficient in the combinatorial model, we adopt particle swarm optimization algorithm (PSO), with the average relative percentage error (MAPE) as the objective function, so that the prediction error of the combined model is minimized. Finally, we empirically analyze the fiscal expenditure sequence from January 2000 to December 2015, and forecast the national fiscal expenditure in 2016. The results show that the combined model is better than the single model. Thus provides the scientific basis for our country's financial expenditure budget.
【学位授予单位】：兰州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F812.45;F224

【参考文献】