基于ARIMA模型和ARIMAX模型的山东省GDP的预测与分析

发布时间：2018-05-06 09:48

本文选题：ARMA模型 + ARIMA模型　；参考：《山东大学》2016年硕士论文

【摘要】：GDP是指一个国家或者地区在一定时期内运用生产要素所生产的全部产品和服务的市场价值。GDP不仅常常被视作一个国家或地区经济状况的一个重要指标,而且也为国家和地区在部署战略方针和制定宏观经济政策上提供了一种参考和依据。与此同时我们可以通过数据来检测制定的宏观经济政策的科学性和有效性。除此之外,GDP的统计数据比较准确、计算重复度小,所以统计起来相对容易。GDP和经济增长率、通货膨胀率及失业率这些主要的宏观经济运行指标都有着非常密切的关系,是最基础性的指标。总之,GDP能够展现出一个国家或地区经济状况的完整图像,帮助判断经济发展的趋势和走向。所以,在目前中国经济高速发展的背景下预测GDP是十分有必要的。本文第一章介绍了课题选择的背景意义及时间序列的现状。在第二章介绍了时间序列预测的相关理论。首先,重点介绍了一元时间序列中ARMA模型和ARIMA模型,然而在现实生活中预测往往是受很多因素影响的,所以接下来介绍了多元时间序列中的ARIMAX模型。在第三章和第四章我们进行了实证分析。以山东省的1975年-2013年GDP数据和第三产业产值为研究对象,基于时间序列分析理论中的ARIMA模型和ARIMAX模型,利用SAS软件,经过数据整理、平稳性检验、模型的识别、参数的估计及模型的检验等步骤后对模型进行拟合。从最后的拟合结果可以看出,预测值和真实值非常吻合,真实值均落在预测值的95%置信区间内,且预测值与真实值的相对误差在2%以内,充分说明模型拟合效果好、预测精度高。通过该实证研究表明ARIMA模型和ARIMAX模型可以应用在实际的工作中进行短期的宏观预测。我们选择第三产业产值作为ARIMA模型的输入变量,主要是想对山东省的产业结构进行分析,并在第五章对山东省的产业结构优化提出了几点意见。本文的创新之处在于,引入第三产业产值作为GDP序列的输入变量,并且运用多元时间序列对数据进行拟合,多元时间序列预测要比单纯的一元时间序列预测要复杂一些。最后通过AIC和SBC准则判断出ARIMAX模型要比ARIMA模型要好一些,所以据此推断,还可以加进多个输入变量,来预测GDP的值可能会更准确一些。
[Abstract]:GDP refers to the market value of all products and services produced by the factors of production in a certain period of time. GDP is not only regarded as an important indicator of the economic situation of a country or region. It also provides a reference and basis for countries and regions to deploy strategic policies and formulate macroeconomic policies. At the same time, we can use data to test the science and effectiveness of macroeconomic policies. In addition, the statistical data of GDP are relatively accurate and the calculation repetition is small, so it is relatively easy to count GDP and the economic growth rate, inflation rate and unemployment rate, which are all closely related to the major macroeconomic operating indicators. Is the most basic indicator. In short, GDP can show a complete picture of the economic situation of a country or region, and help judge the trend and trend of economic development. Therefore, it is necessary to forecast GDP under the background of China's rapid economic development. The first chapter introduces the background significance of topic selection and the present situation of time series. In the second chapter, the theory of time series prediction is introduced. Firstly, the ARMA model and the ARIMA model in univariate time series are introduced. However, in real life, the prediction is often influenced by many factors, so the ARIMAX model in multivariate time series is introduced. In the third and fourth chapters, we carry out empirical analysis. Taking the GDP data of Shandong Province from 1975 to 2013 and the output value of the tertiary industry as the research object, based on the ARIMA model and ARIMAX model in the theory of time series analysis, using SAS software, through data collation, stability test, model identification, Parameter estimation and model test are used to fit the model. From the final fitting results, it can be seen that the predicted value is in good agreement with the real value, the true value falls within the 95% confidence interval of the predicted value, and the relative error between the predicted value and the real value is less than 2%, which fully shows that the model fits well. The prediction accuracy is high. The empirical study shows that ARIMA model and ARIMAX model can be applied to short-term macro prediction in practical work. We choose the output value of the tertiary industry as the input variable of the ARIMA model. We mainly want to analyze the industrial structure of Shandong Province and put forward some suggestions on the optimization of the industrial structure of Shandong Province in the fifth chapter. The innovation of this paper is that the tertiary industry output value is introduced as the input variable of GDP series, and the data are fitted by multivariate time series. The prediction of multivariate time series is more complicated than that of simple monadic time series prediction. Finally, it is found that the ARIMAX model is better than the ARIMA model through the AIC and SBC criteria, so we can infer that it is possible to add more input variables to predict the value of GDP more accurately.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F224;F127

【相似文献】