Models with very low Training gain(FIX?)

[Christian Voll]

Hello all,

I´m currently working on a MaxEnt Model in R for South East Asia.

Because I only had 39 validated presence points I set a buffer (with radius 3km) around each point and filled it with points.

So i got around 100.000 presence points.

I run the models with the original and buffered points and a selected predictior variable set.

However, I get really low training gain (Figure1) with the buffered points.

I checked if this is because of high Collinearity
of the Environment variables. This is not the problem.

However if i run the model with only the 39 presence points I get a high training gain.

Can somebody explain this or has a solution?

Appreciate your help.

Christian

[Cooper Marcus]

What is the HFP variable? What are your results like if you omit it?

The idea of buffering your presence points and filling the buffers with additional points seems odd to me - but I'm far from an expert :)

What pixel size are you using, and what is your total pixel count? (eg. 10km squares, and you have 10M pixels?)

Thanks, Cooper

--
You received this message because you are subscribed to the Google Groups "Maxent" group.
To unsubscribe from this group and stop receiving emails from it, send an email to maxent+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/e9d283b8-30be-45dc-a629-375bb7d7cdf8n%40googlegroups.com.

[Christian Voll]

HFP is the Human Footprint Index

I´m using a 100x100m Grid.

I used different numbers of points in total there were 100,000max for training and test.

I´m  not sure how many there are in total but it is the whole Mekong Region (South East Asia).

The training gain stays very low with omitting.

[Humayun Khattak]

Hello Christi and Cooper Marcus,

I am working on climate change impact on species distribution modelling through MaxEnt. However, I am using this software for the first time and am not too familiar with the tool. 

If you can assist me in using the software properly, I will be grateful to you people