search

NCAR / GPEP

GNU General Public License v3.0
9 stars 6 forks source link

Add cross validation ability for ensemble estimation #18

Open guoqiang-tang opened 1 month ago

guoqiang-tang commented 1 month ago

GPEP now has leave-one-out or k-fold cross validation for spatial regression. The ensemble estimation can also benefit from this ability.

andywood commented 1 month ago

I'm not entirely clear what this is calling for (what is the application?). I submitted a possibly related issue on separation of train / predict modes, which may cover what you're thinking here.