[Biomod-commits] Questions about BIOMOD

Christopher A Walter cwalter at usgs.gov
Tue Jul 21 20:32:37 CEST 2009

We are working on a project testing species distribution models against a 
true population census of shrubs in southwestern Mongolia. In addition, we 
are looking at sampling bias and its effect on predictive ability. We have 
set up sampling regimes in R to represent different strategies and biases 
and are interested in using BIOMOD to model the distributions.

We have two questions: 

1) The first question about BIOMOD deals with the independent explanatory 
and independent response inputs. Since we have the "truth" in the shrub 
census data and are sampling from it, we would like to compare the sample 
against the "truth." In other words, instead of splitting our sample data 
into training and testing sets, the split would be the sample and the 
"truth." I have been reading in the manual about the independent inputs, 
and I am confused as to whether or not this is where we would place the 
"truth" data. I have included an example of the initial.state below: 

> Initial.State(Response = sample[,c(27, 28, 29, 30, 31,32,)], Explanatory 
= sample[,15:26], 
+ IndependantResponse = alldata[,c(27, 28, 29, 30, 31,32,)], 
IndependentExplanatory = alldata[,15:26] 

so the object "sample" is a table of our sample of the census, and the 
object "alldata" is the census dataset. Is this correct? If not, is there 
a way to use the census data set to test the sampled data in the models? 

2) We are running multiple batches in a simulation and errors caused by 
small sample sizes are causing the simulations to crash. We would like to 
allow the modeling to stop but the simulation to loop to the next larger 
sample size and continue. Is there a way to trap the errors and prevent 
the modeling package from stopping all together?


Christopher A. Walter 
Biological Science Technician 
USGS Leetown Science Center 
11649 Leetown Road 
Kearneysville, WV 25430 
Phone: 304 724-4479 

"Its easy not to think when you're not told that much."
    -Justin Pierre
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.r-forge.r-project.org/pipermail/biomod-commits/attachments/20090721/81e56bc1/attachment.htm 

More information about the Biomod-commits mailing list