search

GEO-BON / bon-in-a-box-pipelines

BON in a Box 2.0 - Sampling optimisation and indicator pipelines
MIT License
10 stars 7 forks source link

SDM decision tree to choose best method based on context #142

Open jmlord opened 7 months ago

jmlord commented 7 months ago

Please add criterias in the discussion below.

tpoisot commented 7 months ago

I'm curious to see what it would look like - maybe @frousseu can share a first draft? I'm sure @gottacatchenall would have ideas about it too.

frousseu commented 7 months ago

We need to start thinking more seriously about this, but to be honest I don't have a very clear idea about what it would look like yet. Assuming we only have presence-only data and a way to account for sampling bias, it comes down to is there a best method or is an ensemble method better? For the amount of data needed to model in the first place, I really like the idea of looking at stabilization of coefficients (https://doi.org/10.1016/j.ecolmodel.2020.109283), predicted values or surface area covered with the amount of data.