Visualize Principal Component Analysis

Principal component analysis (PCA) reduces the dimensionality of multivariate data, to two or three that can be visualized graphically with minimal loss of information. fviz_pca() provides ggplot2-based elegant visualization of PCA outputs from: i) prcomp and princomp [in built-in R stats], ii) PCA [in FactoMineR], iii) dudi.pca [in ade4] and epPCA [ExPosition]. Read more: Principal Component Analysis (PCA) in R: Compute, Visualize & Interpret.

fviz_pca_ind(): Graph of individuals
fviz_pca_var(): Graph of variables
fviz_pca_biplot(): Biplot of individuals and variables
fviz_pca(): An alias of fviz_pca_biplot()

Note that, fviz_pca_xxx() functions are wrapper around the core function fviz(), which is also a wrapper around the function ggscatter() [in ggpubr]. Therefore, further arguments, to be passed to the function fviz() and ggscatter(), can be specified in fviz_pca_ind() and fviz_pca_var().

Usage

fviz_pca(X, ...)

fviz_pca_ind(
  X,
  axes = c(1, 2),
  geom = c("point", "text"),
  geom.ind = geom,
  repel = FALSE,
  habillage = "none",
  palette = NULL,
  addEllipses = FALSE,
  col.ind = "black",
  fill.ind = "white",
  col.ind.sup = "blue",
  alpha.ind = 1,
  shape.ind = NULL,
  select.ind = list(name = NULL, cos2 = NULL, contrib = NULL),
  max.points = NULL,
  sample.seed = 123,
  ...
)

fviz_pca_var(
  X,
  axes = c(1, 2),
  geom = c("arrow", "text"),
  geom.var = geom,
  repel = FALSE,
  col.var = "black",
  fill.var = "white",
  alpha.var = 1,
  col.quanti.sup = "blue",
  col.circle = "grey70",
  select.var = list(name = NULL, cos2 = NULL, contrib = NULL),
  ...
)

fviz_pca_biplot(
  X,
  axes = c(1, 2),
  geom = c("point", "text"),
  geom.ind = geom,
  geom.var = c("arrow", "text"),
  col.ind = "black",
  fill.ind = "white",
  col.var = "steelblue",
  fill.var = "white",
  gradient.cols = NULL,
  label = "all",
  invisible = "none",
  repel = FALSE,
  habillage = "none",
  palette = NULL,
  addEllipses = FALSE,
  shape.ind = NULL,
  title = "PCA - Biplot",
  biplot.type = c("auto", "form", "covariance"),
  ...
)

Arguments

X

an object of class PCA [FactoMineR]; prcomp and princomp [stats]; dudi and pca [ade4]; expoOutput/epPCA [ExPosition].

...

Additional arguments.

in fviz_pca_ind() and fviz_pca_var(): Additional arguments are passed to the functions fviz() and ggpubr::ggpar().
in fviz_pca_biplot() and fviz_pca(): Additional arguments are passed to fviz_pca_ind() and fviz_pca_var().

axes

a numeric vector of length 2 specifying the dimensions to be plotted.

geom

a text specifying the geometry to be used for the graph. Allowed values are the combination of c("point", "arrow", "text"). Use "point" (to show only points); "text" to show only labels; c("point", "text") or c("arrow", "text") to show arrows and texts. Using c("arrow", "text") is sensible only for the graph of variables.

geom.ind, geom.var

as geom but for individuals and variables, respectively. Default is geom.ind = c("point", "text), geom.var = c("arrow", "text").

repel

logical; whether to use ggrepel to avoid overplotting text labels. The old jitter argument is kept for backward compatibility and is converted to repel = TRUE with a deprecation warning.

habillage

an optional factor variable for coloring the observations by groups. Default value is "none". If X is a PCA object from FactoMineR package, habillage can also specify the supplementary qualitative variable (by its index or name) to be used for coloring individuals by groups (see ?PCA in FactoMineR).

palette

the color palette to be used for coloring or filling by groups. Allowed values include "grey" for grey color palettes; brewer palettes e.g. "RdBu", "Blues", ...; or custom color palette e.g. c("blue", "red"); and scientific journal palettes from ggsci R package, e.g.: "npg", "aaas", "lancet", "jco", "ucscgb", "uchicago", "simpsons" and "rickandmorty". Can be also a numeric vector of length(groups); in this case a basic color palette is created using the function palette.

addEllipses

logical value. If TRUE, draws ellipses around the individuals when habillage != "none".

col.ind, col.var

color for individuals and variables, respectively. Can be a continuous variable or a factor variable. Possible values also include "cos2", "contrib", "coord", "x", and "y". In this case, the colors for individuals/variables are automatically controlled by their qualities of representation ("cos2"), contributions ("contrib"), coordinates (x^2+y^2, "coord"), x values ("x") or y values ("y"). To use automatic coloring (by cos2, contrib, ....), make sure that habillage ="none".

fill.ind, fill.var

same as col.ind and col.var but for the fill color.

col.ind.sup

color for supplementary individuals

alpha.ind, alpha.var

controls the transparency of individual and variable colors, respectively. The value can vary from 0 (total transparency) to 1 (no transparency). Default value is 1. Possible values also include "cos2", "contrib", "coord", "x", and "y". In this case, the transparency for the individual/variable colors are automatically controlled by their qualities ("cos2"), contributions ("contrib"), coordinates (x^2+y^2, "coord"), x values("x") or y values("y"). To use this, make sure that habillage ="none".

shape.ind

optional factor variable to map the point shape of individuals, independently of their color. This allows colouring individuals by one grouping variable (col.ind/habillage) and shaping them by another (e.g. fviz_pca_ind(res, col.ind = group1, shape.ind = group2)). Default NULL keeps the previous behaviour (shape follows the colour grouping). Use + labs(shape = , colour = ) to rename the legends.

select.ind, select.var

a selection of individuals/variables to be drawn. Allowed values are NULL or a list containing the arguments name, cos2 or contrib:

name: is a character vector containing individuals/variables to be drawn
cos2: if cos2 is in [0, 1], ex: 0.6, then individuals/variables with a cos2 > 0.6 are drawn. if cos2 > 1, ex: 5, then the top 5 individuals/variables with the highest cos2 are drawn.
contrib: if contrib > 1, ex: 5, then the top 5 individuals/variables with the highest contrib are drawn
union: logical. When several of name/cos2/contrib are given, FALSE (default) combines them with AND (each condition further narrows the selection); TRUE combines them with OR (an element is kept if it matches any condition), e.g. named items plus the top-cos2 ones.

max.points

integer or NULL. When the individual / row / column cloud has more than max.points points, a random subset of that many points is drawn so the plot stays readable (usable labels instead of an over-plotted cloud). Only the drawn points/labels are thinned: any ellipse (addEllipses) and the group mean point are still computed on the full data, so a convex / confidence frame is not shrunk or inflated by the draw. When points are coloured/split by a group (e.g. habillage), the draw is stratified so every group keeps at least a minimum number of points and none is decimated. When colour, fill or size is mapped to a continuous metric (e.g. col.ind = "cos2"), its scale and legend are pinned to the full-data range, so a point's colour, fill and size do not depend on how many points are drawn. A message reports how many points are shown. The subset is reproducible (see sample.seed) and does not change the caller's random stream. NULL (default) draws every point.

sample.seed

the random seed used to pick the max.points subset, for a reproducible figure. Ignored when max.points is NULL.

col.quanti.sup

a color for the quantitative supplementary variables.

col.circle

a color for the correlation circle. Used only when X is a PCA output.

gradient.cols

vector of colors to use for n-colour gradient. Allowed values include brewer and ggsci color palettes.

label

a text specifying the elements to be labelled. Default value is "all". Allowed values are "all", "none", or a combination of c("ind", "ind.sup", "quali", "var", "quanti.sup"). "ind" can be used to label only active individuals. "ind.sup" is for supplementary individuals. "quali" is for supplementary qualitative variables. "var" is for active variables. "quanti.sup" is for quantitative supplementary variables.

invisible

a text specifying the elements to be hidden on the plot. Default value is "none". Allowed values are "all", "none", or a combination of c("ind", "ind.sup", "quali", "var", "quanti.sup").

title

the title of the graph

biplot.type

type of biplot scaling for fviz_pca_biplot(). Options are:

"auto" (default): Uses range-based rescaling for visualization
"form": Gabriel form biplot scaling (equivalent to stats::biplot(..., scale = 0)), preserving the score geometry used to compare individuals.
"covariance": Gabriel covariance biplot scaling (equivalent to stats::biplot(..., scale = 1)), preserving the variable covariance geometry. Because both clouds share a single set of axes (unlike stats::biplot()'s dual axes), the individuals can appear compressed relative to the variable arrows; use "form" or "auto" if the individual cloud is the focus.

Note: "form" and "covariance" scaling requires prcomp or princomp objects.

Value

a ggplot

Author

Alboukadel Kassambara alboukadel.kassambara@gmail.com

Examples

# \donttest{
# Principal component analysis
# ++++++++++++++++++++++++++++++
data(iris)
res.pca <- prcomp(iris[, -5],  scale = TRUE)

# Graph of individuals
# +++++++++++++++++++++

# Default plot
# Use repel = TRUE to avoid overplotting (slow if many points)
fviz_pca_ind(res.pca, col.ind = "#00AFBB",
   repel = TRUE)


 
# 1. Control automatically the color of individuals 
   # using the "cos2" or the contributions "contrib"
   # cos2 = the quality of the individuals on the factor map
# 2. To keep only point or text use geom = "point" or geom = "text".
# 3. Change themes using ggtheme: https://www.datanovia.com/learn/data-visualization/ggplot2/themes

fviz_pca_ind(res.pca, col.ind="cos2", geom = "point",
   gradient.cols = c("white", "#2E9FDF", "#FC4E07" ))


# Color individuals by groups, add concentration ellipses
# Change group colors using RColorBrewer color palettes
# Read more: https://www.datanovia.com/learn/data-visualization/ggplot2/colors
# Remove labels: label = "none".
fviz_pca_ind(res.pca, label="none", habillage=iris$Species,
     addEllipses=TRUE, ellipse.level=0.95, palette = "Dark2")

             
     
# Change group colors manually
# Read more: https://www.datanovia.com/learn/data-visualization/ggplot2/colors
fviz_pca_ind(res.pca, label="none", habillage=iris$Species,
     addEllipses=TRUE, ellipse.level=0.95,
     palette = c("#999999", "#E69F00", "#56B4E9"))

      
# Select and visualize some individuals (ind) with select.ind argument.
 # - ind with cos2 >= 0.96: select.ind = list(cos2 = 0.96)
 # - Top 20 ind according to the cos2: select.ind = list(cos2 = 20)
 # - Top 20 contributing individuals: select.ind = list(contrib = 20)
 # - Select ind by names: select.ind = list(name = c("23", "42", "119") )
 
 # Example: Select the top 40 according to the cos2
fviz_pca_ind(res.pca, select.ind = list(cos2 = 40))


 # By default several conditions combine with AND (intersection).
 # Use union = TRUE for OR: keep named individuals PLUS the top-contrib ones.
fviz_pca_ind(res.pca,
             select.ind = list(name = c("23", "42"), contrib = 20, union = TRUE))
#> Selection (union / OR): 20 element(s) kept.


 
# Graph of variables
# ++++++++++++++++++++++++++++
  
# Default plot
fviz_pca_var(res.pca, col.var = "steelblue")

 
# Control variable colors using their contributions
fviz_pca_var(res.pca, col.var = "contrib", 
   gradient.cols = c("white", "blue", "red"),
   ggtheme = theme_minimal())

 
    
# Biplot of individuals and variables
# ++++++++++++++++++++++++++
# Keep only the labels for variables
# Change the color by groups, add ellipses
fviz_pca_biplot(res.pca, label = "var", habillage=iris$Species,
               addEllipses=TRUE, ellipse.level=0.95,
               ggtheme = theme_minimal())


# Biplot scaling modes:
# Form biplot - focus on individual distances
fviz_pca_biplot(res.pca, biplot.type = "form",
               label = "var", habillage = iris$Species)

# Covariance biplot - focus on variable correlations
fviz_pca_biplot(res.pca, biplot.type = "covariance",
               label = "var", habillage = iris$Species)


 # }

Usage

Arguments

Value

See also

Author

Examples