Message-ID: <40e66e0b0903060429w3f69045j76159103e4b85859@mail.gmail.com>
Date: 2009-03-06T12:29:29Z
From: Douglas Bates
Subject: Statistics
In-Reply-To: <22364717.post@talk.nabble.com>
On Thu, Mar 5, 2009 at 7:49 PM, per243 <jose.perezsuarez at csiro.au> wrote:
> How can a non-linear regression to calculate the statistical R-square,
> R-square adjusted, RMSE, VIF??
It is not clear that these statistics are meaningful for a nonlinear
regression model. For example, an R^2 value is meaningful when the
model being fit contains the constant model because it compares the
fit of the current model to the fit of the model y ~ 1. Not all
nonlinear regression models contain the constant model as a submodel.