R/LearnerClustAgnes.R

In mlr3cluster: Cluster Extension for 'mlr3'

#' @title Agglomerative Nesting Clustering Learner
#'
#' @name mlr_learners_clust.agnes
#'
#' @description
#' Agglomerative hierarchical clustering.
#' Calls [cluster::agnes()] from package \CRANpkg{cluster}.
#'
#' The predict method uses [stats::cutree()] which cuts the tree resulting from hierarchical clustering into specified
#' number of groups (see parameter `k`). The default number for `k` is 2.
#'
#' @section Initial parameter values:
#' - `keep.diss`:
#'   - Actual default: `n < 100`, where `n` is the number of observations.
#'   - Adjusted default: `FALSE`.
#'   - Reason for change: Avoid storing the dissimilarity matrix in the model to save memory.
#' - `keep.data`:
#'   - Actual default: `TRUE`.
#'   - Adjusted default: `FALSE`.
#'   - Reason for change: Avoid storing the training data in the model to save memory.
#'
#' @templateVar id clust.agnes
#' @template learner
#'
#' @references
#' `r format_bib("kaufman2009finding")`
#'
#' @export
#' @template seealso_learner
#' @template example
LearnerClustAgnes = R6Class(
  "LearnerClustAgnes",
  inherit = LearnerClust,
  public = list(
    #' @description
    #' Creates a new instance of this [R6][R6::R6Class] class.
    initialize = function() {
      param_set = ps(
        metric = p_fct(c("euclidean", "manhattan"), default = "euclidean", tags = "train"),
        stand = p_lgl(default = FALSE, tags = "train"),
        method = p_fct(
          levels = c("average", "single", "complete", "ward", "weighted", "flexible", "gaverage"),
          default = "average",
          tags = "train"
        ),
        keep.diss = p_lgl(tags = "train"),
        keep.data = p_lgl(default = TRUE, tags = "train"),
        trace.lev = p_int(0L, default = 0L, tags = "train"),
        k = p_int(1L, tags = c("train", "cutree", "predict")),
        par.method = p_uty(
          tags = "train",
          depends = quote(method %in% c("flexible", "gaverage")),
          custom_check = crate(function(x) {
            if (!(test_numeric(x) || test_list(x))) {
              return("`par.method` needs to be a numeric vector")
            }
            if (length(x) %in% c(1L, 3L, 4L)) TRUE else "`par.method` needs be of length 1, 3, or 4"
          })
        )
      )

      param_set$set_values(k = 2L, keep.diss = FALSE, keep.data = FALSE)

      super$initialize(
        id = "clust.agnes",
        feature_types = c("logical", "integer", "numeric"),
        predict_types = "partition",
        param_set = param_set,
        properties = c("hierarchical", "exclusive", "complete"),
        packages = "cluster",
        man = "mlr3cluster::mlr_learners_clust.agnes",
        label = "Agglomerative Nesting"
      )
    }
  ),

  private = list(
    .train = function(task) {
      ps = self$param_set
      m = invoke(
        cluster::agnes,
        x = task$data(),
        diss = FALSE,
        .args = remove_named(ps$get_values(tags = "train"), "k")
      )
      if (self$save_assignments) {
        self$assignments = invoke(stats::cutree, tree = m, .args = ps$get_values(tags = c("train", "cutree")))
      }
      m
    },

    .predict = function(task) {
      pv = self$param_set$get_values(tags = "predict")
      if (pv$k > task$nrow) {
        error_input("`k` needs to be between 1 and %i.", task$nrow)
      }

      warn_prediction_useless(self$id)
      partition = self$assignments %??%
        invoke(stats::cutree, tree = self$model, .args = self$param_set$get_values(tags = c("train", "cutree")))

      PredictionClust$new(task = task, partition = partition)
    }
  )
)

#' @include zzz.R
register_learner("clust.agnes", LearnerClustAgnes)