[R] lme How to validate a model with a validation set and a t est set

Austin, Matt maustin at amgen.com
Wed May 11 20:16:38 CEST 2005


If you used all 28 animals to find the model out of a group of candidate
models, I would have my reservations about this 'validation'.  Any
confidence intervals you get from the final model are bound to be overly
optimistic because you haven't accounted for the degrees of freedom chewed
up during the model fitting/finding process.

Matt Austin
Statistician

Amgen 
One Amgen Center Drive
M/S 24-2-C
Thousand Oaks CA 93021
(805) 447 - 7431

"Today has the fatigue of a Friday and the desperation of a Monday"  -- S.
Pearce 2005


-----Original Message-----
From: r-help-bounces at stat.math.ethz.ch
[mailto:r-help-bounces at stat.math.ethz.ch]On Behalf Of Laurent Fanchon
Sent: Wednesday, May 11, 2005 9:56 AM
To: r-help at stat.math.ethz.ch
Subject: [R] lme How to validate a model with a validation set and a
test set


Dear all,

I am still trying to fit a model for habituation with lme (Dogs are 
trotting on a treadmill once a week during 4 consecutive weeks).
I have maybe found a good one.
In order to validate it, I would like to split my data.
I have 28 dogs, I am thinking to build the model with 20 and use the 
remaining 8 to validate it.

Is it a good method to validate the model?
How can I do this?

I would like to use the data from the first 2 weeks as input and see how 
good the model is to predict the data from the last 2 weeks.
I tried to use predict but this function does not take into account the 
data from the first 2 weeks. I can only get the results for the population.

I would appreciate any suggestions for this

Thank for your help

Laurent

______________________________________________
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




More information about the R-help mailing list