R/Hist.R

In CorReg: Linear Regression Based on Linear Structure Between Variables

#' Histograms with clusters
#' 
#' @param x a vector, matrix or data.frame on which the histogram will be computed and ploted.
#' @param plot if FALSE, the histogram is only computed with no graphical output.
#' @param classes vector of classes to color
#' @param col numeric color
#' @param mode one of c("classical","cumsum","density")
#' @param breaks by default: "Sturges"
#' @param ... further arguments and graphical parameters passed to plot.histogram and thence to title and axis.
#' 
#' @examples
#' x <- c(rnorm(50, 0, 1), rnorm(50, 1, 1))
#' classes <- rep(1:2, each = 50)
#' Hist(x, classes)
#' 
#' @export
Hist <- function(x, classes = NULL, plot = TRUE, col = 2:10, mode = c("classical", "cumsum", "density"), breaks = "Sturges", ...) {
  x = as.numeric(x)
  mode <- match.arg(mode)
  if (is.null(classes)) { # classic histogram
    res = hist(x, plot = plot, breaks = breaks, ...)
  } else { # C'est parti
    if (breaks[1] == "Sturges") {
      xrange = hist(x, plot = FALSE)$breaks
    } else {
      xrange = breaks
      xrange = c(min(x), xrange, max(x))
      # TODO : n'ajouter les extremes que si necessaire
    }
    list_classes = unique(classes)
    hvtots = c()
    for (qui in list_classes) {
      # 2) Have your vectors ready
      vi = x[classes == qui]
      # 3) subset your vectors to be inside xrange
      vi = subset(vi, vi <= max(xrange) & vi >= min(xrange))
      # 4) Now, use hist to compute the counts per interval
      hvi = hist(vi, breaks = xrange, plot = FALSE)$counts
      # hv3 = hist(v3,breaks=xrange,plot=FALSE)$counts
      hvtots = rbind(hvtots, hvi)
    }



    # 5) Finally, Generate a Frequency BarPlot that is equivalent to a Stacked histogram
    # barplot(rbind(hv1,hv2,hv3),col=2:4,names.arg=xrange[-1],space=0,las=1,main=maintitle)
    if (breaks[1] != "Sturges") {
      xrange[1] = paste("<", xrange[2], sep = "")
    npoints = length(xrange)
    xrange[npoints] = paste(">", xrange[npoints - 1], sep = "") # TODO un des deux est une inegalite large + indiquer cela dans la doc de la fonction
    }
    if (mode == "density") {

      # 6) You can also generate a Density Barplot
      Total = colSums(hvtots) #+ hv3
      # barplot(rbind(hv1/Total,hv2/Total,hv3/Total),col=2:4,names.arg=xrange[-1],space=0,las=1)
      barplot(hvtots / Total, names.arg = xrange[-1], space = 0, las = 1, col = col, ...)
      # TODO : ajouter la ligne grise comme dans FlashProcess
      res = list(counts = hvtots / Total)
    } else if (mode == "cumsum") {
      barplot(t(apply(hvtots, 1, cumsum)), names.arg = xrange[-1], space = 0, las = 1, col = col, ...)
      res = list(counts = hvtots)

    } else {
      barplot(hvtots, names.arg = xrange[-1], space = 0, las = 1, col = col, ...)
      res = list(counts = hvtots)
    }

  }

  return(res)
}