mlr集成ClusterR的MiniBatchKmeans时par.vals默认值失效

Let's break down why your default clusters value isn't working and how to fix it:

Root Cause #

You placed the default clusters value in the par.vals argument of makeRLearnerCluster, but that's not how mlr handles required parameter defaults.

In mlr:

• Parameters defined in par.set without a default are treated as required—mlr expects users to explicitly set them, and ignores par.vals for these parameters.
• par.vals is meant for overriding existing default values (from par.set parameters) or passing extra, unregistered parameters to the underlying function.

Solution #

Move the default value directly into the makeIntegerLearnerParam definition for clusters, and remove it from par.vals:

#' @export
makeRLearner.cluster.MiniBatchKmeans = function() {
  makeRLearnerCluster(
    cl = "cluster.MiniBatchKmeans",
    package = "ClusterR",
    par.set = makeParamSet(
      # Add default = 2L directly to the parameter definition
      makeIntegerLearnerParam(id = "clusters", lower = 1L, default = 2L),
      makeIntegerLearnerParam(id = "batch_size", default = 10L, lower = 1L),
      makeIntegerLearnerParam(id = "num_init", default = 1L, lower = 1L),
      makeIntegerLearnerParam(id = "max_iters", default = 100L, lower = 1L),
      makeNumericLearnerParam(id = "init_fraction", default = 1, lower = 0),
      makeDiscreteLearnerParam(id = "initializer", default = "kmeans++", values = c("optimal_init", "quantile_init", "kmeans++", "random")),
      makeIntegerLearnerParam(id = "early_stop_iter", default = 10L, lower = 1L),
      makeLogicalLearnerParam(id = "verbose", default = FALSE, tunable = FALSE),
      makeUntypedLearnerParam(id = "CENTROIDS", default = NULL),
      makeNumericLearnerParam(id = "tol", default = 1e-04, lower = 0),
      makeNumericLearnerParam(id = "tol_optimal_init", default = 0.3, lower = 0),
      makeIntegerLearnerParam(id = "seed", default = 1L)
    ),
    # Remove clusters from par.vals since it now has a default in par.set
    par.vals = list(),
    properties = c("numerics", "prob"),
    name = "MiniBatchKmeans",
    note = "Note",
    short.name = "MBatchKmeans",
    callees = c("MiniBatchKmeans", "predict_MBatchKMeans")
  )
}

Verification #

After making this change:

1. When you create the learner without specifying clusters, getLearnerParVals(lrn) will return $clusters [1] 2 as expected.
2. Calling train(lrn, cluster_task) will automatically pass clusters=2 to the ClusterR function, eliminating the missing argument error.

Quick Note on par.vals Usage #

For future reference, par.vals is best used for:

• Overriding default values defined in par.set (e.g., changing batch_size default from 10 to 20)
• Passing hidden/untunable parameters that you don't want to expose in the learner's parameter set

内容的提问来源于stack exchange，提问作者prasiddhi