Someone brought this up in a workshop -- there's a lot of places that blocked CV can be useful (the Roberts 2017 paper discusses several). It might be useful to support blocking (temporal, phylogenetic) and non-spatial clustering (someone provides a distance matrix, we use a knockoff of spatial_clustering_cv to break it into folds).
Someone brought this up in a workshop -- there's a lot of places that blocked CV can be useful (the Roberts 2017 paper discusses several). It might be useful to support blocking (temporal, phylogenetic) and non-spatial clustering (someone provides a distance matrix, we use a knockoff of spatial_clustering_cv to break it into folds).