Extract models from ENMeval

[Sofia Granados]

Hello guys

I used the package ENMeval to make my models with the followings settings:

seq(1, 5, 0.5),  fc = c("LQH", "QH", "L"),

                       categoricals = NULL, n.bg = 10000, method = "block"

Anyone know if is there any function to extract all the models created?, for example to extract each of the 4 models were performed using a RM: 2 and FC: L (4 because my data partitioning method was "block")

Because I have understood that when you write the function: @prediction [['x number']], the model I obtained is done with every point and not an average of the models made with certain configurations. Is this true?

[Jamie M. Kass]

Sofia,

Yes, you are right that the models in e@models (if e is your ENMeval object) are built using all the points for that combination of settings, and not a model average of all the partition iterations. The results table (e@results) does show you how they did via averages of evaluation stats. We have thought about making a new slot to show the evaluation stats of each partition iteration per model, which wouldn't be so hard to extract while and drop into a slot as ENMeval runs. As there seems to be interest in this, I will talk with Bob about putting this into the next version. Until then, if you're good with code, it wouldn't be that difficult to hack the "tuning.R" script to get what you want. I can try to post some code if you'd like.

Jamie Kass
PhD Candidate
City College, NYC

[Sofia Granados]

Hello Jamie

Thank you for your answer, Yes, I would be so grateful if you try to post some code to try to extract that models.