When publishing, what metrics do you output?

[evan teal]

Hello all, I am reading a paper at the moment which uses MaxEnt to make some SDMs but only reports AUC and Gain. From what I understand, if running MaxEnt with ENMeval (as done in this paper) there are many other metrics that are listed (TSS, AUCdif, Kappa, OR10, COR). In your experiences what metrics do you add to your papers to get a holistic understanding of the models you produce? 

apologies if this is not the best forum for this kind of thing. 

Best,
Evan Teal 

[Bede-Fazekas Ákos]

Hello Evan,

Most of the time, AUC and maxTSS are presented in the publications, but I think it is bad practice. Both AUC and maxTSS are measures developed for presence-absence models, while MaxEnt is not a presence-absence model nor is it a presence-pseudoabsence model (for which AUC and maxTSS might be used with caution) but is a presence-background model. The Boyce index (10.1016/S0304-3800(02)00200-4, 10.1016/j.ecolmodel.2006.05.017) or confidence in positive predictions (10.1016/j.ecolmodel.2024.110667) or any other evaluation measure that prioritize the observed presences over the other parts of the confusion matrix are more useful for MaxEnt models, in my opinion.

Have a nice day,
Ákos
____________
Ákos Bede-Fazekas
Centre for Ecological Research, Hungary

--
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 visit https://groups.google.com/d/msgid/maxent/f2182c9a-b2c8-415b-9946-9587d2ccb4f9n%40googlegroups.com.

[alq...@gmail.com]

Hi Akos

Many Thanks for the advice about the AUC & TSS versus the Boyce index. However, Can you cite a reference that runs a comparison amongst those measures of goodness of fit for Presence only models. I hope it is not too much asking.

Regards

 

                                                      Fraserenvironmental.com

To view this discussion visit https://groups.google.com/d/msgid/maxent/91d81c7c-6b83-4bae-a119-c4555d7da286%40gmail.com.

[Bede-Fazekas Ákos]

Hi,
I have no idea, but I hope the other list members could send you some comparative studies.
Have a nice week,

Ákos
____________
Ákos Bede-Fazekas
Centre for Ecological Research, Hungary

To view this discussion visit https://groups.google.com/d/msgid/maxent/008401dd43b6%2487ea21c0%2497be6540%24%40gmail.com.