Skip to content

Obtaining predicted values for glm() function

3 messages · Jeff Newmiller, DHIMAN BHADRA

DHIMAN BHADRA
# 
Dear all.

I am trying to relate daily deaths in a given city with air pollution
values. The model I am using is as follows:

model <- glm(deaths ~ pollution + ns(time, 13) + ns(temp, 7) + ns(rh, 5) +
dow + holiday, family=quasipoisson, data=city)

Where -
deaths: daily number of deaths
pollution: daily measured levels of air pollution
ns(time, 13): smooth function (i.e. cubic spline) of time with 13 degrees
of freedom
ns(temp, 7): smooth function (i.e. cubic spline) of temperature with 7
degrees of freedom
ns(rh, 5): smooth function (i.e. cubic spline) of humidity with 5 degrees
of freedom
dow: day of the week
holiday: dummy variable to indicate public holidays

Having run this model, how do I obtain predicted values of deaths for any
given set of pollution values? Any suggestions will be much appreciated.

Thanks in advance,
Dhiman Bhadra
Assistant professor.
IIM Ahmedabad.
Jeff Newmiller
# 
Please don't repeat post. In fact, you should definitely read the Posting Guide mentioned at the bottom if this or any other post from the list and follow its advice.

The usual way to predict output of a model in R is to use the predict function appropriate to your model object class. You might find reading the help file ?predict.glm helpful. If not, then you might be needing to study some more basic use cases of R so you can interpret the help files.
---------------------------------------------------------------------------
Jeff Newmiller                        The     .....       .....  Go Live...
DCN:<jdnewmil at dcn.davis.ca.us>        Basics: ##.#.       ##.#.  Live Go...
                                      Live:   OO#.. Dead: OO#..  Playing
Research Engineer (Solar/Batteries            O.O#.       #.O#.  with
/Software/Embedded Controllers)               .OO#.       .OO#.  rocks...1k
--------------------------------------------------------------------------- 
Sent from my phone. Please excuse my brevity.
On August 14, 2014 10:54:17 PM PDT, DHIMAN BHADRA <dhimanbhadra at gmail.com> wrote:
DHIMAN BHADRA
# 
Thanks Jeff for the suggestion - will explore the predict function. So far
I was working on the fitted() function but it seems to generate fitted
(response) values only for predictor values in the original data set, not
for values beyond that.

I never intended to spam the mailing list. I re-posted only when I did not
find my original post in my inbox, which led to think that my query has not
reached the R-family.

Thanks again for your suggestion,
Dhiman


On Fri, Aug 15, 2014 at 11:52 AM, Jeff Newmiller <jdnewmil at dcn.davis.ca.us>
wrote: