There are a few things and hard-coded bits of logic that prevent the experimenter from running on regression tasks. it would also be nice to perhaps add a CLI arg to the experimenter that allows problem types to be selected, in case one is interested in only generating/running problems for a specific problem type.
One known hard-coded bit of logic that needs to change is the final ensembler in the ensemble experiment. It is hardcoded to always be a classifier; it should be dependent on the problem type the pipeline is built for.
There are a few things and hard-coded bits of logic that prevent the experimenter from running on regression tasks. it would also be nice to perhaps add a CLI arg to the experimenter that allows problem types to be selected, in case one is interested in only generating/running problems for a specific problem type.
One known hard-coded bit of logic that needs to change is the final ensembler in the ensemble experiment. It is hardcoded to always be a classifier; it should be dependent on the problem type the pipeline is built for.