Multivariate local polynomial regression for estimating average derivatives

被引：23

作者：

Li, Q ^{[1
]}

Lu, XW

Ullah, A

机构：

[1] Texas A&M Univ, Dept Econ, College Stn, TX 77843 USA

[2] Tsing Hua Univ, Sch Econ & Management, Dept Econ, Beijing 100084, Peoples R China

[3] Univ Calgary, Dept Math & Stat, Calgary, AB T2N 1N4, Canada

[4] Univ Calif Riverside, Dept Econ, Riverside, CA 92521 USA

来源：

JOURNAL OF NONPARAMETRIC STATISTICS | 2003年 / 15卷 / 4-5期

关键词：

average derivatives; local polynomial fitting; kernel method; asymptotic normality;

D O I：

10.1080/10485250310001605450

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

In this paper we suggest to use the sample average of the derivative estimators from a local polynomial fitting to estimate the average derivatives of an unknown multivariate function. Using the techniques of Masry (1996a.b), we derive the asymptotic normal distribution of the proposed average derivative estimator. Monte Carlo experiments show that the proposed estimator compares well with the existing estimators.

引用

页码：607 / 624

页数：18

共 29 条

[1] REGRESSION BY LOCAL FITTING - METHODS, PROPERTIES, AND COMPUTATIONAL ALGORITHMS [J].