K-Fold using pre-defined folds?

2 views
Skip to first unread message

ahmed dawod

unread,
Jul 8, 2016, 12:57:52 PM7/8/16
to eop-users
I'm just wondering is there a possibility to use k-fold training in EOP? I know that the Experimenter class as a feature for k-fold but I viewed its code but I cannot figure out how to pass the folds and rest of the parameters in the command line as I do with the EOPRunner class.

I mean in the ExperimenterCommandOptions there is no '-train' argument. Does this mean I cannot train using this? I can only test?

And I found that I can pass the number of folds using the 'xval' argument but how can I pass the folds themselves. I mean is there a way to pass specific data file as the 1st fold, .... and so on? Or it only splits the data by its own?

Thanks,
Reply all
Reply to author
Forward
0 new messages