R/multcompBoxplot.R

In multcompView: Visualizations of Paired Comparisons

#' boxplot with multcomp graphics
#' 
#' Create boxplots with multcompTs and / or multcompLetters
#' 
#' For formula = y~z, if 'sortFn' is a function or the name of a function,
#' 'multcompBoxplot' starts by applying sortFn to the subsets of y
#' corresponding to each level of z, and then sorting those summaries in
#' increasing or decreasing order, per 'decreasing'.  If 'sortFn' is NULL or
#' NA, this sort step is skipped.
#' 
#' 'multcompBoxplot' then creates 'boxplot' as specified in 'plotList'.  Next,
#' 'compFn' is called to generate comparisons to feed to the functions
#' (\code{\link{multcompTs}} and / or \code{\link{multcompLetters}}, whose
#' output is then passed to (\code{\link{plot.multcomp}}) for plotting.
#' Components of the relevant sublists of 'plotList' are made available to
#' \code{\link{par}} or (for \code{\link{plot.multcompLetters}}) to
#' \code{\link[grid]{gpar}}.
#' 
#' @param formula a two sided formula like "y~z", where both "y" and "z" are
#' columns of the data.frame "data", "y" is numeric, and "z" is a factor.  This
#' will be passed as the first argument for both 'boxplot' and 'compFn', and so
#' must work in both contexts.  NOTE: Any more complicated formula may produce
#' errors or unanticipated results.
#' @param data A data.frame for evaluating 'formula'.
#' @param horizontal TRUE for horizontal boxplots and vertical multcompTs and /
#' or multcompLetters; FALSE for the opposite.
#' @param compFn a function whose output will serve as the the only non-default
#' input to either 'multcompTs' or 'multcompLetters'.  The default "TukeyHSD"
#' actually translates to 'TukeyHSD(aov(formula, data))[[1]][, "p adj"]'.
#' @param sortFn If sortFn is a function or a character string naming a
#' function, it is used to summarize the subset of y corresponding to each
#' level of z into a single number, which will then be used to sort the levels
#' of z according to the argumment 'decreasing'.  This step is skipped if
#' sortFn is NULL or NA or if it is neither a function nor a character string
#' that might name a function.  If sortFn is a character string but a function
#' by that name is not found in the search path, multcompBoxplot stops with,
#' 'Error in do.call(sortFn, list(x = x)) : could not find function ...'.
#' @param decreasing If the levels of z are to be sorted using the output of
#' 'sortFn', this is uses as the 'decreasing' in 'order' to sort the levels of
#' z for plotting.
#' @param plotList A list with names in c("boxplot", "multcompTs",
#' "multcompLetters").  Replicates are allowed.  If present, they produce,
#' e.g., multiple "multcompTs" side by side.  This can be used to compare the
#' visual effects of different arguments to "plot.multcompTs".  Each component
#' of 'plotList' is itself a list of arguments to pass to either "boxplot",
#' "plot.multcompTs" or "plot.multcompLetters".  Placement can be controlled
#' via 'fig' arguments passed (indirectly) of the form 'c(x1, x2, y1, y2)'.
#' If(horizontal==TRUE), fig gives the coordinates of the figure region in the
#' display region of the plot device, as described on the \code{\link{par}}
#' help page; if(horizontal==FALSE), fig[c(3,4,1,2)] gives c(x1, x2, y1, y2)
#' for placement of that portion of the plot.
#' @return This function invisibly returns a list with one component for each
#' component of plotList containing the output of the appropriate
#' "plot.multcomp" call plus the output of "compFn".
#' @author Spencer Graves
#' @seealso \code{\link{boxplot}} \code{\link{multcompTs}}
#' \code{\link{multcompLetters}} \code{\link{plot.multcomp}}
#' \code{\link{TukeyHSD}} \code{\link{par}} \code{\link[grid]{gpar}}
#' @keywords aplot
#' @importFrom stats TukeyHSD aov model.frame terms
#' @importFrom graphics plot
#' @examples
#' 
#' # Example from help("TukeyHSD")
#' multcompBoxplot(breaks~tension, data=warpbreaks)
#' # 'sortFn' can be either a function or a function name
#' # default order is 'decreasing=TRUE'
#' multcompBoxplot(breaks~tension, data=warpbreaks,
#'        sortFn=median, decreasing=FALSE)
#' 
#' ##################
#' library(multcomp)
#' data(recovery)
#' # Horizontal boxplots with both
#' # multcomp Ts and Letters on the right
#' # Using recovery{multcomp} data set
#' multcompBoxplot(minutes~blanket, recovery)
#' 
#' # Plotting boxes rather than letters and Ts
#' 
#' multcompBoxplot(minutes~blanket, recovery,
#'                 plotList=list(
#'                     boxplot=list(fig=c(0, 0.75, 0, 1), las=1,
#'                         cex.axis=1.5),
#'                     multcompTs=list(fig=c(0.7, 0.85, 0, 1),
#'                         type='boxes'),
#'                     multcompLetters=list(
#'                         fig=c(0.87, 0.97, 0.03, 0.98),
#'                         type='boxes') ) )
#' 
#' ####################
#' 
#' # Vertical boxplots with both
#' # multcomp Ts and Letters on the top
#' multcompBoxplot(minutes~blanket, recovery,
#'                         horizontal=FALSE)
#' 
#' # Horizontal boxplots with 2 different
#' # displays of the "Ts" on the left
#' multcompBoxplot(minutes~blanket, recovery,
#'   plotList=list(
#'       boxplot=list(fig=c(0.3, 1, 0, 1)),
#'       multcompTs=list(fig=c(0, 0.15, 0, 1),
#'         orientation="reverse"),
#'       multcompTs=list(fig=c(0.15, 0.3, 0, 1),
#'         type="boxes", orientation="reverse",
#'         mar=c(5,2, 4, 0)+.1) ) )
#' 
#' library(MASS)
#' anorx <-
#' multcompBoxplot(Postwt~Treat, data=anorexia)
#' 
#' \dontrun{
#' # Confirm than sortFn=NULL or NA
#' # leaves the order unchanged
#' library(multcomp)
#' data(cholesterol)
#' cholesterol$trt3 <- with(cholesterol, factor(
#'   as.character(trt), levels=levels(trt)[c(5:4,1:3)]))
#' multcompBoxplot(response ~ trt3, cholesterol,
#'            sortFn=NULL)
#' multcompBoxplot(response ~ trt3, cholesterol,
#'            sortFn=NA)
#' }
#' 
#' @export
"multcompBoxplot" <-
function(formula, data, horizontal=TRUE,
  compFn="TukeyHSD", sortFn="mean",
  decreasing=TRUE, plotList=list(
    boxplot=list(fig=c(0, 0.75, 0, 1)),
    multcompTs=list(fig=c(0.7, 0.85, 0, 1)),
    multcompLetters=list(fig=c(0.87, 0.97, 0.03, 0.98),
      fontsize=20, fontface="bold"))){
##
## 1.  if(!horizontal)swap plotList$...$fig...
##
  n.mc <- length(plotList)
  if(!horizontal)
    for(i in 1:n.mc)
      if("fig" %in% names(plotList[[i]]))
        plotList[[i]]$fig <-
          plotList[[i]]$fig[c(3,4,1,2)]
##
## 2.  Sort the levels of 'z' in the formula
##
  if(!is.null(sortFn) && (is.function(sortFn) ||
       ((!is.na(sortFn)) && is.character(sortFn)))){
    fm <- as.character(formula)
    data <- data[, fm[-1]]
    y.z <- tapply(data[, fm[2]], data[, fm[3]],
           function(x)do.call(sortFn, list(x=x)))
    oz <- order(y.z, decreasing=decreasing)
    data[, fm[3]] <- factor(data[, fm[3]],
           levels=names(y.z)[oz])
  }
##
## 3.  Create the desired boxplots
##
  bpArgs <- plotList$boxplot
  if("fig" %in% names(bpArgs)){
    op <- par(fig=bpArgs$fig)
    on.exit(par(op))
    bpArgs$fig <- NULL
  }
#
  bpArgNames <- names(bpArgs)
  nArgs.bp <- length(bpArgNames)
  bpArg <- vector(3+nArgs.bp, mode="list")
  names(bpArg) <- c("formula", "data",
       "horizontal", bpArgNames)
  bpArg[[1]] <- formula
  bpArg[[2]] <- data
  bpArg[[3]] <- horizontal
  if(nArgs.bp>0)for(i in 1:nArgs.bp)
    bpArg[[3+i]] <- bpArgs[[i]]
  bp <- do.call("boxplot", bpArg)
# Create list "out" and save this.
  out <- vector(n.mc+1, mode="list")
  plotNames <- names(plotList)
  names(out) <- c(plotNames[1], "compFn", plotNames[-1])
  out[[1]] <- bp
  par(op)
##
## 4.  Call "compFn" for the other portions of the
##     display.
##
  Fn <- compFn
  if(compFn=="TukeyHSD"){
    TukeyHSD. <- function(formula, data){
      TukeyHSD(aov(formula, data))[[1]][, "p adj"]
    }
    Fn <- "TukeyHSD."
  }
  if (compFn == "kruskalmc"){
    kruskalmc. <- function(formula, data){
      extract_p(pgirmess::kruskalmc(resp = formula, data = data))
    }
    Fn  <- "kruskalmc."
  }
  fnValue <- do.call(Fn,
          list(formula=formula, data=data))
  Fn.v0 <- vec2mat(fnValue)
  Lvls <- bp$names
  if(horizontal)Lvls <- rev(Lvls)
  FnValue <- Fn.v0[Lvls, Lvls]
  out[["compFn"]] <- fnValue
##
## 5.  Process the remaining components of plotList
##
  if(n.mc>1){
    mar2 <- (horizontal*c(5, 0, 4, 0)+
           (!horizontal)*c(0, 4, 0, 2)+0.1)
    mcNames <- plotNames[-1]
#    plotNms <- paste("plot", mcNames, sep=".")
    if("multcompTs" %in% mcNames)
      mcTs <- multcompTs(FnValue)
    if("multcompLetters" %in% mcNames)
      mcLtrs <- multcompLetters(FnValue)
#   Create the desired multcompViews
    for(i in 1:(n.mc-1)){
      pL.i <- plotList[[i+1]]
      {
        if("mar" %in% names(pL.i)){
          mar <- pL.i$mar
          pL.i$mar <- NULL
        }
        else
          mar <- mar2
      }
      op.i <- par(mar=mar)
      on.exit(par(op.i))
      names.i <- names(pL.i)
      nm.i <- length(names.i)
      args.i <- vector(3+nm.i, mode="list")
      names(args.i) <- c("x", "horizontal",
                         "add", names.i)
      args.i[[1]] <- {
        if(mcNames[i]=="multcompTs")
          mcTs else mcLtrs
      }
      args.i[[2]] <- !horizontal
      args.i[[3]] <- TRUE
      if(nm.i>0)for(ji in 1:nm.i)
        args.i[[3+ji]] <- pL.i[[ji]]
      out[[2+i]] <- do.call(plot, args.i)
      par(op.i)
    }
  }
  invisible(out)
}