plot_PCA: PCA plot using grindr layers

In Momocs: Morphometrics using R

PCA plot using grindr layers

Description

Quickly vizualise PCA objects and friends and build customs plots using the layers. See examples.

Usage

plot_PCA(
  x,
  f = NULL,
  axes = c(1, 2),
  palette = NULL,
  points = TRUE,
  points_transp = 1/4,
  morphospace = TRUE,
  morphospace_position = "range",
  chull = TRUE,
  chullfilled = FALSE,
  labelpoints = FALSE,
  labelgroups = FALSE,
  legend = TRUE,
  title = "",
  center_origin = TRUE,
  zoom = 0.9,
  eigen = TRUE,
  box = TRUE,
  axesnames = TRUE,
  axesvar = TRUE
)

Arguments

x	a PCA object
f	factor specification to feed fac_dispatcher
axes	numeric of length two to select PCs to use (c(1, 2) by default)
palette	color palette to use col_summer by default
points	logical whether to draw this with layer_points
points_transp	numeric to feed layer_points (default:0.25)
morphospace	logical whether to draw this using layer_morphospace_PCA
morphospace_position	to feed layer_morphospace_PCA (default: "range")
chull	logical whether to draw this with layer_chull
chullfilled	logical whether to draw this with layer_chullfilled
labelpoints	logical whether to draw this with layer_labelpoints
labelgroups	logical whether to draw this with layer_labelgroups
legend	logical whether to draw this with layer_legend
title	character if specified, fee layer_title (default to "")
center_origin	logical whether to center origin
zoom	numeric zoom level for the frame (default: 0.9)
eigen	logical whether to draw this using layer_eigen
box	logical whether to draw this using layer_box
axesnames	logical whether to draw this using layer_axesnames
axesvar	logical whether to draw this using layer_axesvar

Value

a plot

Note

This approach will replace plot.PCA (and plot.lda in further versions. This is part of grindr approach that may be packaged at some point. All comments are welcome.

See Also

Other grindr: drawers, layers_morphospace, layers, mosaic_engine(), papers, pile(), plot_LDA(), plot_NMDS()

Examples

### First prepare two PCA objects.

# Some outlines with bot
bp <- bot %>% mutate(fake=sample(letters[1:5], 40, replace=TRUE)) %>%
efourier(6) %>% PCA
plot_PCA(bp)
plot_PCA(bp, ~type)
plot_PCA(bp, ~fake)

# Some curves with olea
op <- olea %>%
mutate(s=coo_area(.)) %>%
filter(var != "Cypre") %>%
chop(~view) %>% opoly(5, nb.pts=90) %>%
combine %>% PCA
op$fac$s %<>% as.character() %>% as.numeric()

op %>% plot_PCA(title="hi there!")

### Now we can play with layers
# and for instance build a custom plot
# it should start with plot_PCA()

my_plot <- function(x, ...){

x %>%
    plot_PCA(...) %>%
    layer_points %>%
    layer_ellipsesaxes %>%
    layer_rug
}

# and even continue after this function
op %>% my_plot(~var, axes=c(1, 3)) %>%
    layer_title("hi there!")

# grindr allows (almost nice) tricks like highlighting:

# bp %>% .layerize_PCA(~fake) %>%
#   layer_frame %>% layer_axes() %>%
#   layer_morphospace_PCA() -> x

# highlight <- function(x, ..., col_F="#CCCCCC", col_T="#FC8D62FF"){
#  args <- list(...)
#  x$colors_groups <- c(col_F, col_T)
#  x$colors_rows <- c(col_F, col_T)[(x$f %in% args)+1]
#  x
#  }
# x %>% highlight("a", "b") %>% layer_points()

# You get the idea.

Momocs documentation built on Nov. 13, 2023, 5:07 p.m.