R/summary.clustatis.R

In ClustBlock: Clustering of Datasets

## =============================================================================


##' @title Show the CLUSTATIS results
##'
##' @usage
##' \method{summary}{clustatis}(object, ngroups=NULL, ...)
##'
##' @description
##' This function shows the clustatis results
##'
##'
##' @param object object of class 'clustatis'.
##'
##' @param ngroups number of groups to consider. Ignored for clustatis_kmeans results. Default: recommended  number of clusters
##'
##' @param ... further arguments passed to or from other methods
##'
##'
##' @return the CLUSTATIS principal results
##'
##'
##'
##' @keywords quantitative
##'
##' @return a list with:
##'         \itemize{
##'          \item group: the clustering partition
##'          \item homogeneity: homogeneity index (%) of each cluster and the overall homogeneity index (%) of the partition
##'          \item weights: weight associated with each block in its cluster
##'          \item rho: the threshold for the noise cluster
##'           \item test_one_cluster: decision and pvalue to know if there is more than one cluster
##'          }
##'
##'
##' @seealso   \code{\link{clustatis}} , \code{\link{clustatis_kmeans}}
##'
##' @export


## =============================================================================





summary.clustatis <- function(object, ngroups = NULL, ...) {
  res.clustatis <- object
  if (inherits(res.clustatis, "clustatis") == FALSE) {
    stop("The class of the object must be 'clustatis'")
  }

  if (is.null(ngroups) | res.clustatis$type == "K") {
    ngroups <- res.clustatis$param$ng
  }

  if (res.clustatis$type == "H+C") {
    if (ngroups > res.clustatis$param$gpmax) {
      stop("ngroups>gpmax")
    }
  }


  if (res.clustatis$type == "H+C") {
    test_one_cluster <- res.clustatis$test_one_cluster
    res.clustatis <- res.clustatis[[ngroups]]
  } else {
    test_one_cluster <- "No test"
  }
  NameBlocks <- rownames(res.clustatis$group)

  liste_groups <- list()
  for (i in 1:ngroups)
  {
    liste_groups[[i]] <- NameBlocks[res.clustatis$group == i]
  }
  names(liste_groups) <- paste("Cluster", 1:ngroups)
  if (sum(res.clustatis$rho) > 0) {
    liste_groups[[ngroups + 1]] <- NameBlocks[res.clustatis$group == "K+1"]
    names(liste_groups)[ngroups + 1] <- "Noise cluster (K+1)"
  }

  res <- list(
    groups = liste_groups, homogeneity = res.clustatis$homogeneity,
    weights = res.clustatis$weights, rho = res.clustatis$rho
  )


  return(res)
}