The blockCV package creates spatially or environmentally separated training and testing folds for cross-validation to provide a robust error estimation in spatially structured environments. See
The tidymodels group at RStudio is working on adding some spatial methods to our infrastructure, starting with resampling methods. (https://www.tmwr.org and https://www.tidymodels.org/start/resampling are good primers if you've never seen anything on tidymodels).
We already have a PR for spatial resampling using clustering (tidymodels/rsample#185) and would probably create one using blockCV. Since you have a sf dependency, we probably wouldn't add that to rsample.
Would you be interested in having those tidymodels functions reside in blockCV? It would mean an rsample dependency.
rvalavi
commented
2 years ago
The tidymodels group at RStudio is working on adding some spatial methods to our infrastructure, starting with resampling methods. (https://www.tmwr.org and https://www.tidymodels.org/start/resampling are good primers if you've never seen anything on tidymodels).
We already have a PR for spatial resampling using clustering (tidymodels/rsample#185) and would probably create one using
blockCV. Since you have asfdependency, we probably wouldn't add that torsample.Would you be interested in having those tidymodels functions reside in
blockCV? It would mean anrsampledependency.