基于滑动窗口的区间值部分可加线性模型的宏观经济预测  

Macroeconomic Forecasting of Interval-valued Partially Additive Linear Model Based on Sliding Window

作　　者：陶志富[1] 杨金 刘兮 Tao Zhifu;Yang Jin;Liu Xi(School of Economics,Anhui University,Hefei 230601,China;School of Mathematics and Statistics,Hefei Normal University,Hefei 238076,China)

机构地区：[1]安徽大学经济学院,合肥230601 [2]合肥师范学院数学与统计学院,合肥238076

出　　处：《统计与决策》2022年第19期21-25,共5页Statistics & Decision

基　　金：国家自然科学基金青年项目(72001001,71901088);安徽省高校人文社会科学研究项目(SK2020A0049);安徽省自然科学基金青年项目(1808085QG221)。

摘　　要：针对区间型数据,文章在部分线性模型的基础上结合可加模型并引入滑动窗口模型,提出了基于滑动窗口的中点、极差部分可加线性模型,融合了半参数回归和滑动窗口模型的优点,同时又避免了维数灾难。依据交叉熵准则确定了滑动窗口的期数,并基于最小二乘法及核估计方法给出了模型参数和未知函数估计的迭代算法。在实证分析中,通过引入若干金融指标,对宏观经济进行预测。结果表明,改进模型优于传统回归模型。For interval data,this paper combines the additive model with the sliding window model based on the partial linear model,and proposes a partially additive linear model of the midpoint and range based on the sliding window model,which combines the advantages of semi-parametric regression and sliding window model,and avoids dimension disaster.Then,the paper accords to the cross-entropy criterion to determine the period number of the sliding window,and is also based on the least square method and kernel estimation method to offer the iterative algorithm for estimating the model parameters and unknown functions.Finally,in the empirical analysis,the paper brings in a number of financial indicators to forecast the macro economy.The results show that the improved model is better than the traditional regression model.

关 键 词：区间型数据 滑动窗口模型 部分可加线性模型 交叉熵 宏观经济预测 

分 类 号：F224.9[经济管理—国民经济]