search

sylvainschmitt / SSDM

Stacked Species Distribution Modelling R package
Other
41 stars 17 forks source link

Evaluation metrics for future projection #113

Closed AJ-KBA closed 3 years ago

AJ-KBA commented 3 years ago

I would like to know how the SSDM function 'project' is able to return evaluation metrics such as 'AUC', 'prop.correct' when there are no occurrences (train/test) for future scenarios?

lukasbaumbach commented 3 years ago

The evaluation metrics refer to the original model you used for projection. The project function gives you the option to A) just return the projected (future scenario) raster (output.format="rasters") or B) return the original model with exchanged raster in the @projection slot (output.format="model") Also see the documentation for project.

AJ-KBA commented 3 years ago

The evaluation metrics refer to the original model you used for projection. The project function gives you the option to A) just return the projected (future scenario) raster (output.format="rasters") or B) return the original model with exchanged raster in the @projection slot (output.format="model") Also see the documentation for project.

Thank You. This was helpful.