EVOLUTION-MANAGER
Edit File: gam.scale.html
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><title>R: Scale parameter estimation in GAMs</title> <meta http-equiv="Content-Type" content="text/html; charset=utf-8" /> <link rel="stylesheet" type="text/css" href="R.css" /> </head><body> <table width="100%" summary="page for gam.scale {mgcv}"><tr><td>gam.scale {mgcv}</td><td style="text-align: right;">R Documentation</td></tr></table> <h2>Scale parameter estimation in GAMs</h2> <h3>Description</h3> <p>Scale parameter estimation in <code><a href="gam.html">gam</a></code> depends on the type of <code>family</code>. For extended families then the RE/ML estimate is used. For conventional exponential families, estimated by the default outer iteration, the scale estimator can be controlled using argument <code>scale.est</code> in <code><a href="gam.control.html">gam.control</a></code>. The options are <code>"fletcher"</code> (default), <code>"pearson"</code> or <code>"deviance"</code>. The Pearson estimator is the (weighted) sum of squares of the pearson residuals, divided by the effective residual degrees of freedom. The Fletcher (2012) estimator is an improved version of the Pearson estimator. The deviance estimator simply substitutes deviance residuals for Pearson residuals. </p> <p>Usually the Pearson estimator is recommended for GLMs, since it is asymptotically unbiased. However, it can also be unstable at finite sample sizes, if a few Pearson residuals are very large. For example, a very low Poisson mean with a non zero count can give a huge Pearson residual, even though the deviance residual is much more modest. The Fletcher (2012) estimator is designed to reduce these problems. </p> <p>For performance iteration the Pearson estimator is always used. </p> <p><code><a href="gamm.html">gamm</a></code> uses the estimate of the scale parameter from the underlying call to <code>lme</code>. <code><a href="bam.html">bam</a></code> uses the REML estimator if the method is <code>"fREML"</code>. Otherwise the estimator is a Pearson estimator. </p> <h3>Author(s)</h3> <p> Simon N. Wood <a href="mailto:simon.wood@r-project.org">simon.wood@r-project.org</a> with help from Mark Bravington and David Peel</p> <h3>References</h3> <p>Fletcher, David J. (2012) Estimating overdispersion when fitting a generalized linear model to sparse data. Biometrika 99(1), 230-237. </p> <h3>See Also</h3> <p><code><a href="gam.control.html">gam.control</a> </code> </p> <hr /><div style="text-align: center;">[Package <em>mgcv</em> version 1.8-28 <a href="00Index.html">Index</a>]</div> </body></html>