R/ggbiplot.R

In low-decarie/ggpca: Creates ggplot biplot methods for pca objects

#' Generic ggplot biplot function
#'
#' @description Produces a ggplot biplot from an PCA analysis similar to biplot.  Currently has methods for princomp and prcomp.
#' @param PC a fitted object
#' @param selected.pc vector of length 2 number of two principle components
#' @param groups vector group assigned to each point
#' @param length.alpha logical wether transparency of explanatory variables should vary with length
#' @param scale scaling for explanatory variable arrows
#' @param varnames
#' @param ... other arguments to be passed to qplot
#' @details If \code{groups} acts as stats:::biplot
#' If \code{groups} is specified and has 6 or less unique values, observations are presented as different shapes for each group (ggplot \code{shape} argument) and circled with a 95% confidence interval ellipses, each group having an ellipsis with a different \code{linetype}.
#' If \code{groups} is specified and has more than 6 unique values, observations are presented with different colours for each group (ggplot \code{colour} argument) and circled with a 95% confidence interval ellipses, each group having an ellipsis with a different \code{colour}.
#' 
#' @return a ggplot object
#' @keywords pca
#' @seealso \code{\link{biplot}}, \code{\link{prcomp}}, \code{\link{princomp}}
#' @export
#' @examples
#'# Can be used on the results of princomp and prcomp
#'p <- ggbiplot(PC=princomp(iris[,-5]))
#'print(p)
#'# For 6 or less groups, uses shape and linetype
#'p <- ggbiplot(PC=prcomp(iris[,-5]),
#'                    selected.pc=c(1,2),
#'                    groups=iris[,5],
#'                    main="PCA of iris dataset")
#'print(p)
#'# For more than 6 groups, uses colour
#'p <- ggbiplot(PC=prcomp(iris[,-5]),
#'                    selected.pc=c(1,2),
#'                    groups=sample(LETTERS[1:7],nrow(iris), replace=T),
#'                    main="PCA of iris dataset")
#'print(p)

ggbiplot <- function(PC,
                     selected.pc=c(1,2),
                     groups=NULL,
                     scale=1,
                     length.alpha=T,
                     ...) UseMethod("ggbiplot", PC)