StackedEnsemble should allow user to pass in metalearner type

Description

  • Currently the metalearner is hardcoded as a default H2O GLM with non-negative weights. (`non_negative = TRUE`)

  • We need to add a `metalearner_algorithm` argument to allow customization of the metalearning algorithm.
    Initially, this argument will take in either "glm" (default), "gbm", "drf", or "deeplearning".

  • A user cannot specify the metalearner model hyperparameters at this point. This will be completed with the addition of another argument, e.g. `metalearner_params`

  • junit, runit, pyunit for this argument is included

Assignee

Navdeep

Fix versions

Reporter

Erin LeDell

Support ticket URL

None

Labels

None

Affected Spark version

None

Customer Request Type

None

Task progress

None

CustomerVisible

Yes

Epic Link

Components

Priority

Major
Configure