pca: Principal Components Analysis (PCA) with NA (missing data)

In vanderleidebastiani/SYNCSA: Analysis of Functional and Phylogenetic Patterns in Metacommunities

Description

The function use the option "pairwise.complete.obs" (in function cor) for calculate the correlation. The correlation between each pair of variables is computed using all complete pairs of observations on those variables.

Usage

pca(data)

## S3 method for class 'pcasyncsa'
plot(
  x,
  show = c("variables", "individuals"),
  axis = c(1, 2),
  xlab = axis[1],
  ylab = axis[2],
  arrows = TRUE,
  text = TRUE,
  points = FALSE,
  ...
)

Arguments

data	A data frame or matrix with individuals in rows and variables in columns.
x	A object of class pcasyncsa.
show	Draw "variables" or "individuals".
axis	Axis for draw, must have length equal to two (Default axis = c(1, 2)).
xlab	Text for x label (Default xlab = axis[1]).
ylab	Text for y label (Default ylab = axis[2]).
arrows	Logical argument (TRUE or FALSE) to specify if arrows are showed for variables (Default arrows = TRUE).
text	Logical argument (TRUE or FALSE) to specify if text are showed for individuals (Default text = TRUE).
points	Logical argument (TRUE or FALSE) to specify if points are showed for individuals (Default points = FALSE).
...	Parameters for plot function.

Value

decomposition	list with the results of decomposition of correlation matrix.
eigenvalues	Data frame containing all the eigenvalues, the percentage of inertia and the cumulative percentage of inertia.
individuals	Coordinates for the individuals.
variables	Correlation between original variables and axes.

Author(s)

Vanderlei Julio Debastiani <[e-mail]>

See Also

syncsa syncsa

Examples

data(ADRS)
traits<-ADRS$traits
# Some NA
traits[c(1,5),1]<-NA
traits[3,2]<-NA
traits
res<-pca(traits)
res
plot(res, show = "variables", arrows = TRUE)
plot(res, show = "individuals", axis = c(1, 2), text = TRUE)
plot(res, show = "individuals", text = FALSE, points = TRUE)

vanderleidebastiani/SYNCSA documentation built on Sept. 4, 2020, 10:57 p.m.