Say if I want to do (one time) training and test set cross validation - any refs for optimal % either way? Vaguely recall 10-20% training...
-
yes I've seen those args for k folds but looking for 1 test equivalent for reasons that don't fit into a tweet :)
I see! :) Usually people just do 50-50 split. But it depends on how data hungry your model is. E.g. wouldn't do GWAS with 50/50.
-
-
Indeed, that's my consideration. I've seen a case made for smaller exploration set though which would be nice
-
Need larger explrtn set to prvent overfitting. Use precise betas from stp1 to create cmpsit for 1df test in stp2.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.