On Wed, 2011-09-28 at 02:10 -0700, pigpigmeow wrote:

> For example:

> GAMs and after stepwise regression:

You probably don't want to be doing stepwise model/feature selection in

any regression model. Marra & Wood (2011, Computational Statistics and

Data Analysis 55; 2372-2387) show results that suggest adding an extra

penalty term during model fitting, that allows terms to be penalised out

of the model, performs well for feature selection in a series of GAM

exercises.

You can turn this on in mgcv::gam() via the `select` argument by setting

it to `TRUE`.

> cod<-gam(newCO~RH+s(solar,bs="cr")+windspeed+s(transport,bs="cr"),family=gaussian

> (link=log),groupD,methods=REML)

>

> I used 10 year meterorology data (2000-2010) to form equation of

> concentration of carbon monoxide.

> NOW, I have 2011 meteorology data, I want to use the above GAMs to get the

> predict value of concentration of carbon monoxide.

So put your 2011 data into a data frame with (at the minimum) the column

names:

`RH`, `solar`, `windspeed`, `transport`

If that data frame were called, say, `dat2011` then you provide the

`predict()` method for "gam" objects with both the fitted model **and**

the new data (`dat2011`), e.g.:

pred <- predict(cod, newdata = dat2011)

See ?predict.gam for more details. You are probably going to need type =

"response" so you get values back on the scale of the response not the

the link function.

HTH

G

--

%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%

Dr. Gavin Simpson [t] +44 (0)20 7679 0522

ECRC, UCL Geography, [f] +44 (0)20 7679 0565

Pearson Building, [e] gavin.simpsonATNOSPAMucl.ac.uk

Gower Street, London [w]

http://www.ucl.ac.uk/~ucfagls/ UK. WC1E 6BT. [w]

http://www.freshwaters.org.uk%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%

______________________________________________

[hidden email] mailing list

https://stat.ethz.ch/mailman/listinfo/r-helpPLEASE do read the posting guide

http://www.R-project.org/posting-guide.htmland provide commented, minimal, self-contained, reproducible code.