Linearly constrained local polynomial regression /
Local polynomial regression is a popular nonparametric regression technique. In some cases, additional information is available about the regression function (or its derivatives) that one may wish to incorporate into the estimator. I shall address essentials of local polynomial regression, how one...
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
[Place of publication not identified] :
[publisher not identified] ;
2001.
|
| Subjects: | |
| Online Access: | http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=726102681&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD |
| Summary: | Local polynomial regression is a popular nonparametric regression technique. In some cases, additional information is available about the regression function (or its derivatives) that one may wish to incorporate into the estimator. I shall address essentials of local polynomial regression, how one may incorporate linear equality constraints into a local polynomial regression estimate, and how such constraints may impact the bias and variance of the regression estimator. Tools for automatically selecting the bandwidth of the local polynomial smoother will be suggested. A specific application of linear equality constraints in local polynomial regression will be addressed: estimation of a symmetric regression function. In addition, I will use quadrature to construct linear constraints for cases when information about integrals of the regression function are known. I will apply this technique to density estimation settings for copula density functions and for univariate density estimation with known quantiles. In a topic related to local polynomial regression, I will develop methods for estimating a homoskedastic variance in a nonparametric regression problem when there are multiple covariates. |
|---|---|
| Item Description: | Vita. "Major Subject: Statistics". |
| Physical Description: | xvii, 202 leaves : illustrations ; 28 cm. Issued also on microfiche from University Microfilm Inc. |
| Bibliography: | Includes bibliographical references (leaves 186-191). |