PCA Loadings Biplot — pca_biplot • TempleCBE

A classic PCA biplot: the observation scores (pca_model$x) overlaid with variable loading vectors (pca_model$rotation) drawn as labeled arrows from the origin. Observations can optionally be colored by a categorical grouping variable with 95% concentration ellipses.

Usage

pca_biplot(
  pca_model,
  x = 1,
  y = 2,
  group = NULL,
  ellipse = FALSE,
  title = "PCA Biplot",
  percent = FALSE
)

Arguments

pca_model: A prcomp object or cbe_pca object from run_pca.
x, y: Which principal components to plot on the x/y axes (default 1, 2).
group: Optional categorical vector (e.g. iris$Species) to color observations by.
ellipse: Logical (default FALSE); if TRUE and group is provided, draws 95% confidence/concentration ellipses around each group.
title: Optional plot title (default "PCA Biplot").
percent: Logical (default FALSE); if TRUE, appends percent variance explained to axis labels.

Value

A ggplot object.

Details

Loading vectors are unit-scale by construction and would be invisible next to the score cloud if plotted as-is, so they are rescaled so that their maximum extent is 80% of the score cloud's maximum extent (a standard biplot convention) before being drawn.

Examples

pca_model <- prcomp(iris[, 1:4], center = TRUE, scale. = TRUE)
pca_biplot(pca_model, group = iris$Species, ellipse = TRUE)