If you need an accessible version of this item please contact JSTOR User Support

Bayesian "Confidence Intervals" for the Cross-Validated Smoothing Spline

Grace Wahba
Journal of the Royal Statistical Society. Series B (Methodological)
Vol. 45, No. 1 (1983), pp. 133-150
Published by: Wiley for the Royal Statistical Society
Stable URL: http://www.jstor.org/stable/2345632
Page Count: 18
  • Download PDF
  • Cite this Item

You are not currently logged in.

Access your personal account or get JSTOR access through your library or other institution:

login

Log in to your personal account or through your institution.

If you need an accessible version of this item please contact JSTOR User Support
Bayesian
Preview not available

Abstract

We consider the model Y(ti) = g(ti) = εi, i = 1, 2,..., n, where g(t), t ∈ [ 0, 1] is a smooth function and the {εi} are independent N(0, σ2) errors with σ2 unknown. The cross-validated smoothing spline can be used to estimate g non-parametrically from observations on Y(ti), i = 1, 2,..., n, and the purpose of this paper is to study confidence intervals for this estimate. Properties of smoothing splines as Bayes estimates are used to derive confidence intervals based on the posterior covariance functiion of the estimate. A small Monte Carlo study with the cubic smoothing spline is carried out to suggest by example to what extent the resulting 95 per cent confidence intervals can be expected to cover about 95 per cent of the true (but in practice unknown) values of g(ti), i = 1, 2,..., n. The method was also applied to one example of a two-dimensional thin plate smoothing spline. An asymptotic theoretical argument is presented to explain why the method can be expected to work on fixed smooth functions (like those tried), which are "smoother" than the sample functions from the prior distributions on which the confidence interval theory is based.

Page Thumbnails

  • Thumbnail: Page 
[133]
    [133]
  • Thumbnail: Page 
134
    134
  • Thumbnail: Page 
135
    135
  • Thumbnail: Page 
136
    136
  • Thumbnail: Page 
137
    137
  • Thumbnail: Page 
138
    138
  • Thumbnail: Page 
139
    139
  • Thumbnail: Page 
140
    140
  • Thumbnail: Page 
141
    141
  • Thumbnail: Page 
142
    142
  • Thumbnail: Page 
143
    143
  • Thumbnail: Page 
144
    144
  • Thumbnail: Page 
145
    145
  • Thumbnail: Page 
146
    146
  • Thumbnail: Page 
147
    147
  • Thumbnail: Page 
148
    148
  • Thumbnail: Page 
149
    149
  • Thumbnail: Page 
150
    150