R/buildSummary.R

In Crossover: Analysis and Search of Crossover Designs

#' Build Summary Table For All Examples From Literature
#' 
#' Build Summary Table For All Examples From Literature
#' 
#' See also the documentation for the data files.
#' 
#' @param extended If \code{TRUE} the summary table will have further 
#' columns with extended information as how balanced the design is and
#' whether all treatment effect differences are estimable under all
#' models.
#' @return TODO
#' @author Kornelius Rohmeyer \email{rohmeyer@@small-projects.de}
#' @references See the documentation for the data files.
#' @keywords misc
#' @examples
#' 
#'   buildSummaryTable()
#' 
#' @export buildSummaryTable
buildSummaryTable <- function(extended=FALSE) {	
	summaryTable <- data.frame(dataset=character(0), title=character(0), reference=character(0), signature=character(0), t=numeric(0), p=numeric(0), s=numeric(0))
	path <- system.file("data", package="Crossover")
	for (file in dir(path=path)) {	
    # clatworthy1.rda contains the PBIB(2) designs while clatworthyC.rda / pbib2combine.rda contains the crossover designs.
    if (file %in% c("clatworthy1.rda", "exampleSearchResults2t.rda")) next
		designs <- load(paste(path, file, sep="/"), envir=Crossover.env)
		for (design in designs) {
			dataset <- design
			design <- get(design, envir=Crossover.env)
			title <- attr(design, "title")
			reference <- attr(design, "reference")
			signature <- attr(design, "signature")
			p <- dim(design)[1]
			s <- dim(design)[2]
			t <- length(levels(as.factor(design)))
      if (!extended) {
        summaryTable <- rbind(summaryTable, 
                              data.frame(dataset=dataset, title=title, reference=reference, signature=signature, t=t, p=p, s=s))
      } else {
        variances <- general.carryover(design, model=1)$Var.trt.pair[upper.tri(diag(t))]        
        balance.delta <- max(variances)-min(variances)
        
        # The following line is for the CRAN warning...
        est1 <- est2 <- est3 <- est4 <- est5 <- est6 <- est7 <- est8 <- 0
        for (i in 1:8) {
          assign(paste("est", i, sep=""), estimable(design, t, i))
        }
        
        summaryTable <- rbind(summaryTable, 
                              data.frame(dataset=dataset, title=title, reference=reference, signature=signature, t=t, p=p, s=s,
                                         balance.delta=balance.delta, est1, est2, est3, est4, est5, est6, est7, est8))
      }
		}
	}
	summaryTable$dataset <- as.character(summaryTable$dataset)
	summaryTable$title <- as.character(summaryTable$title)
	summaryTable$signature <- as.character(summaryTable$signature)
	summaryTable$reference <- as.character(summaryTable$reference)
	return(summaryTable)
}

getSignatureStr <- function(design) {
	p <- dim(design)[1]
	s <- dim(design)[2]
	t <- length(levels(as.factor(design)))
	return(paste("p=",p,", s=",s,",t=",t,sep=""))
}

getSignature <- function(design) {
	p <- dim(design)[1]
	s <- dim(design)[2]
	t <- length(levels(as.factor(design)))
	return(c(p,s,t))
}

#getTable <- function(design) {
#    design <- design2integer(design)#
#	rownames(design) <- paste("p", 1:dim(design)[1], sep="")
	#colnames(design) <- paste("s", 1:dim(design)[2], sep="")
#	return(paste(capture.output(print(xtable(design), type="html")), collapse="\n"))
#}

design2integer <- function(design) {
    return(matrix(as.integer(design), dim(design)[1]))
}

getEff <- function(design, ppp=0.5, placebos=1) {
  average.eff <- rep(0, 9)
  for (model in 1:9) {
	  eff <- design.efficiency(design, model=model,  model.param=list(ppp=0.5, placebos=1))$eff.trt.pair.adj
    average.eff[model] <- mean(eff[row(eff)!=col(eff)])
  }
	return(average.eff)
}

CrossoverVersion <- function() {
	x <- try(as.character(packageVersion("Crossover")), silent=TRUE)
	if (!is(class(x), "try-error")) {
		return(x)
	} else {
		return("unknown")
	}
}

getOffDiagMean <- function(x) {
  return(mean(x[row(x)!=col(x)]))
}