基于Copula函数的CPI相关性分析及模糊预测

发布时间：2018-01-10 15:19

本文关键词：基于Copula函数的CPI相关性分析及模糊预测　出处：《宁夏大学》2017年硕士论文　论文类型：学位论文

【摘要】：CPI是重要的经济指标,反映着居民消费水平的变动情况.国家的宏观经济调控政策的出台力度将受到CPI大小的直接影响.PPI是衡量企业生产成本变化的指标,也是经济核算的重要依据.对这两个指标的定量分析是当今统计学研究的热门问题之一,本文一方面基于Copula函数研究CPI与PPI间的相关性度量,这有利于分析经济过热或紧缩的发展趋势.另一方面,我们提出了一类模糊回归组合预测模型,并将其运用于CPI的短期预测研究,为制定经济政策提供可靠的依据和建议,从而减少因为政策的时滞性带来的影响.一类经济数据呈现出显著的非线性关系,我们常采用的Pearson相关系数并不能够正确、客观地反映非线性性.Copula作为刻画变量间相关关系的工具,在非线性情况下具有显著的优势.它可以灵活、全面、精细地刻画出变量间复杂的相依性结构.本文考虑到CPI与PPI的本质非线性相关,基于表现能力更强的Copula来研究CPI与PPI之间的相依性.采用2011年1月至2016年12月全国月度同比CPI与PPI数据进行相关性研究,结果表明Gaussian Copula在刻画秩相关性上效果较好,而在尾部相关性以及与原始数据的拟合程度上,GumbelCopula要优于其他Copula.通常由于输入或输出数据的不精确造成了经典回归模型的局限性,模糊回归利用模糊集理论在刻画不精确性的优势,有效的解决了这一问题.在预测中,单纯地使用一种模型难以全面揭示其变化规律.组合预测理论能有效利用单一模型的有用信息提高预测准确度.本文提出了一种模糊回归组合预测方法,并运用到我国CPI的预测中.该方法能综合模糊回归和组合预测的优点,给出更符合实际的CPI预测区间值.最后,CPI预测结果表明,模糊回归组合模型的预测结果优于各单项预测模型,能提高CPI的预测精度,适合对CPI进行短期预测.
[Abstract]:CPI is an important economic indicator. Reflecting the change of consumption level of residents. The national macroeconomic regulation and control policy will be directly affected by the size of CPI. PPI is an index to measure the change of enterprise production cost. The quantitative analysis of these two indexes is one of the hot issues in the statistical research. On the one hand, based on the Copula function, this paper studies the correlation measure between CPI and PPI, which is helpful to analyze the development trend of economic overheating or contraction. On the other hand. We put forward a kind of fuzzy regression combination forecasting model and apply it to the short-term forecast research of CPI, which provides reliable basis and suggestion for making economic policy. In order to reduce the impact of policy delay, a class of economic data show a significant nonlinear relationship, we often use the Pearson correlation coefficient can not be correct. As a tool to describe the correlation between variables, the objective reflection of nonlinearity. Copula has significant advantages in nonlinear cases. It can be flexible and comprehensive. The complex structure of dependence between variables is described in detail. In this paper, the essential nonlinear correlation between CPI and PPI is considered. The dependency between CPI and PPI was studied based on the more expressive Copula. The correlation between CPI and PPI data from January 2011 to December 2016 was used to study the correlation between CPI and PPI. Research. The results show that Gaussian Copula is effective in describing rank correlation, but in tail correlation and fitting degree with original data. GumbelCopula is superior to other Copula. The limitations of classical regression models are usually due to inaccuracy of input or output data. Fuzzy regression utilizes the advantage of fuzzy set theory to depict imprecision and solve this problem effectively. It is difficult to reveal the law of change by using a single model alone. The combination prediction theory can effectively use the useful information of a single model to improve the prediction accuracy. In this paper, a fuzzy regression combined prediction method is proposed. This method can integrate the advantages of fuzzy regression and combination prediction, and give the actual CPI prediction interval value. Finally, the results show that the method can be used to predict CPI. The prediction result of fuzzy regression combination model is better than that of each single prediction model, which can improve the prediction accuracy of CPI, and is suitable for short-term prediction of CPI.
【学位授予单位】：宁夏大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224;F124

【参考文献】