Re: Input script for model selection with ENMTools

[Robert Long]

Here is some additional information/clarification points to my original post:

I'm hoping to use ENMTools with MaxEnt to evaluate the support among a number of models. I have 7-8 environmental layers that I want to include, and also the potential for many many model alternatives depending on which feature combinations I choose. My question has to do with the input script required by ENMTools for model selection, which appears to only allow one ascii per line (i.e., per analysis). But, a given model in MaxEnt can (and often does) use multiple ascii files—one for each “environmental layer” included in that run. So, for example, three of the models I'd like to compare might be:

1. Environmental layers: elevation; Features: Linear only
2. Environmental layers: elevation, slope; Features: Linear only
3. Environmental layers: elevation, slope; Features: Linear, quadratic

The first is straightforward...the input script would have a single ascii file for "elevation." But, what would the input rows look like in the script for the 2nd and 3rd models? Can you specify more than one ascii per model?

Thanks for any thoughts...maybe I'm just misunderstanding something.


On Saturday, September 29, 2012 7:13:08 PM UTC-7, Robert Long wrote:

Hi,

I'm attempting to use ENMTools to conduct model selection across a number of potential MaxEnt models. It seems that the script required by ENMTools can have only one ascii file listed per line (i.e., per analysis). But, what I’m having trouble reconciling is that a given analysis in MaxEnt can (and often does) use multiple ascii files—one for each “environmental layer” included in that run. Is the model selection component in ENMTools only designed to help choose between models vis-a-vis the number and types of "features" (e.g., linear, quad, product, threshold, hinge) included various models, but only on models with a single environmental layer? This seems fairly limiting, so maybe I'm missing something? Can anyone offer any thoughts?

Thanks,

Robert

[david p]

The ascii file that you need is the final Maxent output from your model (not the different ascii input layers); i.e., run all of your different models with your variables using Maxent, and write your script to compare the final output from each different model using ENMTools. Maxent ran the models, ENM tools only compares the final output.

Cheers,

David

[Paul Jensen]

Hi Robert,


Another thing to be careful of when using model selection is to make
sure that your data are fixed among models. In other words if you have
some variables with missing values, your sample sizes among models will
vary (i.e., you can't compare a model with n = 100 vs. a model with n =
50). It's probably not a major problem if sample sizes vary only
slightly, but could be if your sample sizes were quite different. It's
easy to overlook this point when comparing a lot of models.


Best,


Paul

>>> david p 10/01/12 5:17 PM >>>

--
You received this message because you are subscribed to the Google
Groups "Maxent" group.
To view this discussion on the web visit
https://groups.google.com/d/msg/maxent/-/cPj6FW3t8WMJ.
To post to this group, send email to max...@googlegroups.com.
To unsubscribe from this group, send email to
maxent+un...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/maxent?hl=en.

[Robert Long]

Thanks, David. This is exactly the type of answer I was hoping for!

Best,

Robert

[yasmin...@students.mq.edu.au]

Hi David,

sowats about the point CSV file, should i include the points of each species used for each model or its somthing resulted from Maxent ouputs. Also, is it only the long and lat coloumns without species name coloumn?

cheers

yasmin