Code snippet in R to produce clustering results through parallel coordinates

plotlabels.R

# labels is a NxL data.frame where N=number of observations being clustered and
# L = number of clustering results (number of vertical lines)

plot.label.distribution <- function(labels,
                                    groupColumn,
                                    alphaLines = 1/10,
                                    useSplines = T,
                                    showPoints = T,
                                    scale='uniminmax', ...) {

  library(GGally)
  library(ggplot2)

  N <- nrow(labels)
  L <- ncol(labels)

  if (missing(groupColumn)) groupColumn <- L
  labels[,(L+1):(2*L)] <- lapply(labels, function(l)jitter(as.numeric(l))) #add noise
  colnames(labels)[(L+1):(2*L)] <- as.character(seq_len(L))

  ggp <- ggparcoord(labels, L+seq(L), scale=scale, groupColumn=groupColumn,
    alphaLines = alphaLines,
    showPoints = showPoints,
#    useSplines = useSplines,
    ...) +
    xlab('Number of clusters') +
    ylab('Observation memberships') +
    guides(colour = F) +
    theme_minimal() +
    theme(axis.ticks = element_blank(), axis.text.y = element_blank())

#  if(useSplines) ggp <- ggp + scale_x_continuous(breaks=seq(L))
  ggp
}