Research

Coverage properties of confidence intervals for generalized additive model components


Reference:

Marra, G. and Wood, S., 2011. Coverage properties of confidence intervals for generalized additive model components. Scandinavian Journal of Statistics, 39 (1), pp. 53-74.

Related documents:

[img]
Preview
PDF (MarraWoodCI.pdf) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Download (540kB) | Preview

    Official URL:

    http://dx.doi.org/10.1111/j.1467-9469.2011.00760.x

    Abstract

    We study the coverage properties of Bayesian confidence intervals for the smooth component functions of generalized additive models (GAMs) represented using any penalized regression spline approach. The intervals are the usual generalization of the intervals first proposed by Wahba and Silverman in 1983 and 1985, respectively, to the GAM component context. We present simulation evidence showing these intervals have close to nominal ‘across-the-function’ frequentist coverage probabilities, except when the truth is close to a straight line/plane function. We extend the argument introduced by Nychka in 1988 for univariate smoothing splines to explain these results. The theoretical argument suggests that close to nominal coverage probabilities can be achieved, provided that heavy oversmoothing is avoided, so that the bias is not too large a proportion of the sampling variability. Otherwise, because the Bayesian intervals account for bias and variance, the coverage probabilities are surprisingly insensitive to the exact choice of smoothing parameter. The theoretical results allow us to derive alternative intervals from a purely frequentist point of view, and to explain the impact that the neglect of smoothing parameter variability has on confidence interval performance. They also suggest switching the target of inference for component-wise intervals away from smooth components in the space of the GAM identifiability constraints. Instead intervals should be produced for each function as if only the other model terms were subject to identifiability constraints. If this is done then coverage probabilities are improved.

    Details

    Item Type Articles
    CreatorsMarra, G.and Wood, S.
    DOI10.1111/j.1467-9469.2011.00760.x
    DepartmentsFaculty of Science > Mathematical Sciences
    Publisher StatementMarraWoodCI.pdf: The definitive version is available at onlinelibrary.wiley.com
    RefereedYes
    StatusPublished
    ID Code26662

    Export

    Actions (login required)

    View Item

    Document Downloads

    More statistics for this item...