SDM workflow
All SDM approaches have a similar generic workflow:
Conceptualisation
Make a plan for your work first.
Data
The butterflies of Pakistan and maybe of whole Asia - depending on the outcome of the digitization part of this course - will serve as example in this course.
Model fitting
There are plenty of methods available for this step. Here are the arguably most important ones, which will be introduced by your tutorials:
| Method |
|---|
| Profile methods |
| Bioclim |
| Domain |
| Regression methods |
| Generalized linear models |
| Generalized additive models |
| Machine-learning methods |
| MaxEnt |
| Random forest |
| Boosted regression trees |
| Neural networks |
| Support vector machines |
| XG Boost |
| Bayesian nonparametric regression |
| Ensemble methods |
| Model averaging |
Assessment
Depends a bit on the method, but not too much.
Predictions
Are easy to make but difficult to reproduce and even more difficult to evaluate.