From brodgar@lists.highstat.com Mon Dec 8 09:55:30 2008
From: brodgar@lists.highstat.com (brodgar@lists.highstat.com)
Date: Mon, 8 Dec 2008 09:55:30 +0000
Subject: [brodgar]Brodgar question
Message-ID: <260DC8CDC4E86A4EB4E3CE6D9600401C075F4D@geesemail.adf.bham.ac.uk>
Hello,
I am currently working with GAM in Brodgar and I have encountered a =
problem. Could anybody give any suggestions?
=20
I am trying to understand the impact of coal mine drainage (water and =
sediments polluted with iron and other components) on species richness =
on several Welsh streams. I have ecological and environmental data =
measured up and downstream from each mine and I want to see the =
relationship between the changes in the response variable and the =
explanatory variables.
After simplifying my list of environmental variables (eliminating =
correlated variables, those that do not change downstream from the mines =
and samples with outliers), I have 21 samples.=20
Iron in sediments seems to be linearly related to species richness, but =
the other variables (especially iron in water) have a clear non-linear =
relationship.=20
I have tried to apply a GAM into the data in order to find these =
relationships in Brodgar 2.5.6, but if I try to put more than 1 smoother =
at a time it crashes, giving the following error message:
Error in smooth.construct.tp.smooth.spec(object, data, knots) :=20
A term has fewer unique covariate combinations than specified maximum =
degrees of freedom
Calls: source ... smooth.construct -> smooth.construct.tp.smooth.spec -> =
.C
Execution halted
=20
Any of the environmental variables used has a lot of identical values.
If I enter the smoothers one at a time it works. The same occurs if I =
use GAMM.
=20
Does anybody have any suggestions to why this is happening? Is it =
because I do not have enough samples? Any thoughts are appreciated.
=20
Thank you
=20
Montserrat Auladell
From brodgar@lists.highstat.com Mon Dec 8 10:55:52 2008
From: brodgar@lists.highstat.com (brodgar@lists.highstat.com)
Date: Mon, 08 Dec 2008 10:55:52 +0000
Subject: [brodgar](no subject)
Message-ID: <20081208105556.9AC2D103800E@smtpauth03.csee.onr.siteprotect.com>
have tried to apply a GAM into the data in order to find these
=
relationships in Brodgar 2.5.6, but if I try to put more than 1 smoother
=
at a time it crashes, giving the following error message:
Error in smooth.construct.tp.smooth.spec(object, data, knots) :=20
A term has fewer unique covariate combinations than specified
maximum =
degrees of freedom
Calls: source ... smooth.construct -> smooth.construct.tp.smooth.spec
-> =
.C
Execution halted
=20
Any of the environmental variables used has a lot of identical values.
This problem is indeed due to a covariate not having enough unique
values. A solution is to avoid using cross-validation (see the updated
online manuals how to do this). Set it to 3 or 4. But 21 observations is
not much! I wouldn't use more than 3 or 4 explanatory variables for such
a small data set. And don't do GAMMs with spatial correlations.
Good luck.
Alain
Dr. Alain F. Zuur
First author of:
1. Analysing Ecological Data (2007).
Zuur, AF, Ieno, EN and Smith, GM. Springer. 680 p.
URL:
www.springer.com/0-387-45967-7
2. Mixed effects models and extensions in ecology with R. (2008).
Zuur, AF, Ieno, EN, Walker, N, Saveliev, AA, and Smith, GM.
Springer.
http://www.springer.com/life+sci/ecology/book/978-0-387-87457-9
Other books:
http://www.brodgar.com/books.htm
Statistical consultancy, courses, data analysis and software
Highland Statistics Ltd.
6 Laverock road
UK - AB41 6FN Newburgh
Tel: 0044 1358 788177
Email: highstat@highstat.com
URL:
www.highstat.com
URL:
www.brodgar.com