R help - Jan 2006 - negative predicted values in poisson glm

Dear  R  helpers,
running the following code of a glm model of the family poisson, gives
predicted values < 0. Why?

library(MASS)
library(stats)
library(mvtnorm)
library(pscl)
data(bioChemists)
poisson_glm <-  glm(art ~ fem + mar + kid5 + phd + ment, data = bioChemists,
family = poisson)
predicted.values = predict(poisson_glm)
range(predicted.values)


Thank you in advance for any hints.
Best regards,
P. Olsson

	[[alternative HTML version deleted]]

On Wed, 18 Jan 2006, P. Olsson wrote:
> Dear  R  helpers,
> running the following code of a glm model of the family poisson, gives
> predicted values < 0. Why?
Look at the help page for predict.glm, particularly the "type"
argument.
Then exponentiate the results.

 	-thomas

>
> library(MASS)
> library(stats)
> library(mvtnorm)
> library(pscl)
> data(bioChemists)
> poisson_glm <-  glm(art ~ fem + mar + kid5 + phd + ment, data =
bioChemists,
> family = poisson)
> predicted.values = predict(poisson_glm)
> range(predicted.values)
>
>
> Thank you in advance for any hints.
> Best regards,
> P. Olsson
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html
>
Thomas Lumley			Assoc. Professor, Biostatistics
tlumley at u.washington.edu	University of Washington, Seattle

"P. Olsson" <olsson1 at gmail.com> writes:
> Dear  R  helpers,
> running the following code of a glm model of the family poisson, gives
> predicted values < 0. Why?
> 
> library(MASS)
> library(stats)
> library(mvtnorm)
> library(pscl)
> data(bioChemists)
> poisson_glm <-  glm(art ~ fem + mar + kid5 + phd + ment, data =
bioChemists,
> family = poisson)
> predicted.values = predict(poisson_glm)
> range(predicted.values)
> 
> 
> Thank you in advance for any hints.

The prediction is on the link scale.


-- 
   O__  ---- Peter Dalgaard             ??ster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark          Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)                  FAX: (+45) 35327907

P. Olsson <olsson1 <at> gmail.com> writes:
> 
> Dear  R  helpers,
> running the following code of a glm model of the family poisson, gives
> predicted values < 0. Why?
> 
  Because by default predict.glm() gives answers on the 
link (log) scale.  predict(poisson_glm,type="response")
is probably what you're looking for.  See ?predict.glm.

  Ben Bolker

On Wed, 18 Jan 2006 14:58:26 +0100, P. Olsson wrote:

PO> Dear  R  helpers,
PO> running the following code of a glm model of the family poisson,
PO> gives predicted values < 0. Why?
PO> 
PO> library(MASS)
PO> library(stats)
PO> library(mvtnorm)
PO> library(pscl)
PO> data(bioChemists)
PO> poisson_glm <-  glm(art ~ fem + mar + kid5 + phd + ment, data PO>
bioChemists, family = poisson)
PO> predicted.values = predict(poisson_glm)
PO> range(predicted.values)
PO> 


use
 predicted.values = predict(poisson_glm, type="response")

best wishes,

Adelchi Azzalini  <azzalini at stat.unipd.it>
Dipart.Scienze Statistiche, Universit?? di Padova, Italia
tel. +39 049 8274147,  http://azzalini.stat.unipd.it/