[1]苏梽芳,周煜,李气芳.采用最小二乘支持向量机的部分相依函数型线性模型估计与应用[J].华侨大学学报(自然科学版),2022,43(4):544-552.[doi:10.11830/ISSN.1000-5013.202108037]
　SU Zhifang,ZHOU Yu,LI Qifang.Estimate and Application of Partial Dependent Functional Linear Model Using Least Squares Support Vector Machine[J].Journal of Huaqiao University(Natural Science),2022,43(4):544-552.[doi:10.11830/ISSN.1000-5013.202108037]
点击复制

采用最小二乘支持向量机的部分相依函数型线性模型估计与应用()
分享到：

《华侨大学学报（自然科学版）》[ISSN:1000-5013/CN:35-1079/N]

卷:
第43卷
期数:
2022年第4期
页码:
544-552
栏目:
出版日期:
2022-07-18

文章信息/Info

Title:
Estimate and Application of Partial Dependent Functional Linear Model Using Least Squares Support Vector Machine
文章编号:
1000-5013(2022)04-0544-09
作者:
苏梽芳1;  周煜1;  李气芳2
1. 华侨大学 经济与金融学院, 福建 泉州 362021;2. 闽南师范大学 数学与统计学院, 福建 漳州 363000
Author(s):
SU Zhifang1;  ZHOU Yu1;  LI Qifang2
1. School of Economics and Finance, Huaqiao University, Quanzhou 362021, China; 2. School of Mathematics and Statistics, Minnan Normal University, Zhangzhou 363000, China
关键词:
部分相依函数型线性模型;  长期协方差函数;  相依函数型数据;  最小二乘支持向量机
Keywords:
partial dependent function linear model;  long-term variance function;  dependent functional datasleast;  least squares support vector machine
分类号:
O212
DOI:
10.11830/ISSN.1000-5013.202108037
文献标志码:
A
摘要:
提出一种基于无截断Bartlett核函数的重构方法,有效避免长期方差函数估计方法面临的核函数与窗宽选择问题,并将其应用到部分相依函数型线性模型中.利用考虑函数型数据相依性的最小二乘支持向量机对模型进行参数估计,数值模拟结果表明:与未考虑函数型数据相依特征的最小二乘估计方法相比,提出的考虑函数型数据相依性的最小二乘支持向量机估计方法能更稳健地估计向量系数,有效提高样本外的预测精度;将部分相依函数型线性模型应用到上证指数开盘价的预测中,得到较好的预测效果.
Abstract:
We propose the reconstruction method based on non truncated Bartlett kernel function, in which the selection of kernel function and window width faced by the long-term variance function estimation method are avoided effectively, and apply it to the partial dependent function linear model. The least squares support vector machine considering the dependence of function data is used to estimate the parameters of the model. Numerical simulation results show that, compared with the least squares estimation method not considering the dependent features of functional data, the least squares support vector machine estimation method considering the dependence of functional data is more robustand effectively improve the out-of-sample prediction accuracy. The partial dependent function linear model is applied to the prediction of the opening price of Shanghai stock index and a better prediction effect is obtained.

参考文献/References:

[1] RAMSAY J O,SILVERMAN B W.Functionaldata analysis[M].2nd ed.New York:Springer-Verlag,2005.DOI:10.1002/0470013192.bsa239.
[2] HYNDMAN J,ULLAH S.Robust forecasting of mortality and fertility rates: A functional data approach[J].Computational Statistics and Data Analysis,2007,51(10):4942-4956.DOI:10.1016/j.csda.2006.07.028.
[3] PARK J Y,QIAN Junhui.Functional regression of continuous state distributions[J].Journal of Econometrics,2012,167(2):397-412.DOI:10.1016/j.jeconom.2011.09.024.
[4] 王国华.中国股票市场日内波动率研究[D].武汉:中南财经政法大学,2017.DOI:10.1016/S0167-7152(99)00036-X.
[5] 苏梽芳,傅一铮,包浩华.国债期限利差对经济周期转换的预警指示作用: 基于函数型数据Logit模型的实证分析[J].宏观经济研究,2021(5):20-30.DOI:10.16304/j.cnki.11-3952/f.2021.05.003.
[6] CARDOT H,FERRATY F,SARDA P.Functional linear model[J].Statistics and Probability Letters,1999,45(1):11-22.DOI:10.1016/S0167-7152(99)00036-X.
[7] CARDOT H,FERRATY F,SARDA P.Spline estimators for the functional linear model[J].Statistica Sinica,2003,13(3):571-591.DOI:10.1016/S0266-8920(03)00029-8.
[8] YAO Fang,MüLLER H G,WANG J L.Functional data analysis for sparse longitudinal data[J].Journal of the American Statistical Association,2005,100(470):577-590.DOI:10.1198/016214504000001745.
[9] CRAMBES C,KNEIP A,SARDA P.Smoothing splines estimators for functional linear regression[J].The Annals of Statistics,2009,37(1):35-72.DOI: 10. 1214/07-AOS563.
[10] HALL D P.Methodology and theory for partial least squares applied to functional data[J].The Annals of Statistics,2012,40:322-352.DOI: 10.2307/41713637.
[11] KALOGRIDIS I,AELST S V.Robust functional regression based on principal components[J].Journal of Multivariate Analysis,2019,173:393-415.DOI:10.1016/j.jmva.2019.04.003.
[12] CRAMBES C,KNEIP A,SARDA P.Smoothing splines estimators for functional linear regression[J].The Annals of Statistics,2009,37(1):35-72.DOI:10.1214/07-AOS563.
[13] ZHANG Daowen,LIN Xihong,SOWERS M F.Two-stage functional mixed models for evaluating the effect of longitudinal covariate profiles on a scalar outcome[J].Biometrics,2007,63(2):351-362.DOI:10.1111/j.1541-0420.2006.00713.x.
[14] SHIN H.Partial functional linear regression[J].Journal of Statistical Planning and Inference,2009,139(10):3405-3418.DOI:10.1016/j.jspi.2009.03.001.
[15] ZHOU Jianjun,CHEN Min.Spline estimators for semi-functional linear model[J].Statistics and Probability Letters,2012,82(3):505-513.DOI:10.1016/j.spl.2011.11.027.
[16] 王晓光,高黎.一类部分函数型线性模型的估计方法[J].高校应用数学学报:A辑,2013,28(3):253-265.DOI:10.3969/j.issn.1000-4424.2013.03.001.
[17] H?RMANN S,KOKOSZKA P.Weakly dependent functional data[J].The Annals of Statistics,2010,38(3):1845-1884.DOI:10.1214/09-aos768.
[18] KOKOSZKA P,REIMHERR M.Introduction to functional data analysis[M].New York:CRC Press,2017.
[19] 李气芳.相依函数型数据分析方法及其金融应用[D].厦门:华侨大学,2020.DOI:10.27155/d.cnki.ghqiu.2020.000517.
[20] KIEFER N M,VOGELSANG T J.Heteroskedasticity-autocorrelation robust standard errors using the Bartlett kernel without truncation[J].Econometrica,2002,70(5):2093-2095.DOI:10.1111/1468-0262.00366.
[21] 孟银凤,梁吉业.基于最小二乘支持向量机的函数型数据回归分析[J].模式识别与人工智能,2014,27(12):1124-1130.DOI:10.3969/j.issn.1003-6059.2014.12.009.
[22] H?RMANN S,KIDZISKI L,HALLIN M.Dynamic functional principal components[J].Journal of the Royal Statistical Society: Series B(Statistical Methodology),2015,77(2):319-348.DOI:10.1111/rssb.12076.
[23] KOKOSZKA P,REIMHERR M.Introduction to functional data analysis[M].Boca Raton:CRC Press,2017.

备注/Memo

备注/Memo:
收稿日期: 2021-08-30
通信作者: 苏梽芳(1977-),男,教授,博士,博士生导师,主要从事数量经济模型、函数型数据分析方法的研究.E-mail:suzufine@hqu.edu.cn.
基金项目: 国家社科基金资助项目(21AJY001)http://www.hdxb.hqu.edu.cn
更新日期/Last Update: 2022-07-20