Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

How can I validate my modeling for future periods? #348

Closed
Oliveira-D-R opened this issue Apr 21, 2021 · 4 comments
Closed

How can I validate my modeling for future periods? #348

Oliveira-D-R opened this issue Apr 21, 2021 · 4 comments

Comments

@Oliveira-D-R
Copy link

I ask because the AUC values made available in the modeling process are only for the modeling of the present time.

@jamiemkass
Copy link
Member

Validating future projections, or projections in the present for areas without occurrence data, is difficult because you lack the occurrence data you need to calculate performance metrics that require them, like AUC. Other kinds of validation might include inspecting the prediction for environmental values outside those used for model training with MESS, for example, to determine areas of uncertainty. I guess a model that results in a prediction with big areas that have very negative MESS values might be considered "less valid" than another model that does not, but we might be stretching the term "validation" here.

@hannahlowens
Copy link

+1 to @jamiemkass. MESS (or MOP in KUENM) will show you in which areas your model predictions are most likely to fail. If you have enough occurrence data to do so, you could also try hindcasting the model and seeing how well it predicts occurrences at a past time point. Then you at least have a bit of an idea of how robust your model is through time.

@Oliveira-D-R
Copy link
Author

Oliveira-D-R commented Apr 26, 2021 via email

@gepinillab
Copy link
Member

To run MESS, you first need to project your model (in area or time). Then, you just need to use the last module "Calculate Environmental Similarity" in the Project Component (or Tab).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants