search

mlr-org / mlr3tuning

Hyperparameter optimization package of the mlr3 ecosystem
https://mlr3tuning.mlr-org.com/
GNU Lesser General Public License v3.0
54 stars 5 forks source link

learner API: new function predict_trace #50

Open berndbischl opened 5 years ago

berndbischl commented 5 years ago

many learners have a "sequence" parameter so you can get full trace from a single training run.

e.g. ntree, boosting rounds, or the "s" param in glmnet

this is should be supported.

basically we need to allow to "mark up" one hyperparam in the set, with a tag. and then maybe not have a new predict_trace function but handle this in the normal predict

jakob-r commented 5 years ago

Proposal: tag the Param with "sequential"

berndbischl commented 5 years ago

Proposal: tag the Param with "sequential"

that is not very precise, and debatable whether the "s" param in glmnet falls under it tag = "predict_trace" or "predict_sequential" ?

jakob-r commented 5 years ago

Then the Param would have the tags "train" , "predict" and "predict_sequential", right?

mb706 commented 5 years ago

I think the current "predict" tag is actually enough, but we would have to adjust tuning to handle predict-time-only parameters efficiently

berndbischl commented 5 years ago

I think the current "predict" tag is actually enough, but we would have to adjust tuning to handle predict-time-only parameters efficiently

yes, michel and me discussed this a bit. we need an issue in mlr3tuning now to documents the current thoughts