Hello. I'm getting troble with interpret the jackkife of MaxEnt.

I read a training manual―A training manual for habitat suitability and connectivity modeling using tigers (Panthera tigris) in Bhutan as example (2015)―and in the paper, set the random test percentage to zero, then check the AUC and remove some environmental variables which the "without variable" value is bigger than "with all variables" value. I wanted to try this method. I unchecked the random seed, changed the random test percentage to 0, replicates 10, replicated run type crossvalidate as the paper does. Then I got "Jackknife of regularized training gain for [species]" and "Jackknife of test gain for [species]. I can't understand what does the test gain mean. I set the test percentage to zero, so I expected to get training gain without test gain.

Could you explain me the meaning of the test value in my case?

And I knew that the the replicated run type can be selected depending on the number of presence data(samples). Then can I use bootstrap in this process?

More Seulki Koo's questions See All
Similar questions and discussions