R/getClusters.R

In tscR: A time series clustering package combining slope and Frechet distances

#' Generic clustering function
#'
#' Compute clustering with pam function and a distance class object.
#'
#'
#' @param x Numeric distance object obtained with dimension n x n.
#' @param k Numeric. Number of clusters
#'
#' @details This function is a wrapper of pam (cluster). x must be a
#' dist object obtained from frechetdist, slopedist
#'  or any other distance metric on condition
#'  that it be an object of the dist class and has dimensions nxn,
#'  where n is equal to the number of trajectories.
#'
#' @return Object of class 'pam'. See \code{\link[cluster]{pam.object}}
#'  for details
#'
#'
#' @examples
#'
#' data(tscR)
#' data <- tscR
#' time <- c(1,2,3)
#' dist_tscR <- slopeDist(data, time)
#' res.cluster <- getClusters(dist_tscR, 3)
#'
#'
#' @seealso \code{\link[cluster]{pam}, \link{plotCluster}.}
#'
#' @author  Fernando Pérez-Sanz (\email{fernando.perez8@@um.es})
#' @author  Miriam Riquelme-Pérez (\email{miriam.riquelmep@@gmail.com})



getClusters <- function(x, k) {
  if (!inherits(x, "dist")) {
    stop("X must be a dist class matrix")
  }
  if (length(x) > 1e+06) {
    stop("Large distance could cause memory problems.
             You should use imputeSenators function")
  }
  myclust <- pam(x, k = k, diss = TRUE)
  return(myclust)
}