Skip to contents

Plot FPCA Results

Source: R/fdata.R
plot.fdata2pc.Rd

Visualize functional principal component analysis results with multiple plot types: component perturbation plots, variance explained (scree plot), or score plots.

Usage

# S3 method for class 'fdata2pc'
plot(
  x,
  type = c("components", "variance", "scores"),
  ncomp = 3,
  multiple = 2,
  show_both_directions = TRUE,
  ...
)

Arguments

x

An object of class 'fdata2pc' from fdata2pc.

type

Type of plot: "components" (default) shows mean +/- scaled PC loadings, "variance" shows a scree plot of variance explained, "scores" shows PC1 vs PC2 scatter plot of observations.

ncomp

Number of components to display (default 3 or fewer if not available).

multiple

Factor for scaling PC perturbations. Default is 2 (shows +/- 2*sqrt(eigenvalue)*PC).

show_both_directions

Logical. If TRUE (default), show both positive and negative perturbations (mean + PC and mean - PC). If FALSE, only show positive perturbation. All curves are solid lines differentiated by color.

...

Additional arguments passed to plotting functions.

Value

A ggplot object (invisibly).

Details

The "components" plot shows the mean function (black) with perturbations in the direction of each principal component. The perturbation is computed as: mean +/- multiple * sqrt(variance_explained) * PC_loading. All lines are solid and differentiated by color only.

The "variance" plot shows a scree plot with the proportion of variance explained by each component as a bar chart.

The "scores" plot shows a scatter plot of observations in PC space, typically PC1 vs PC2.

See also

fdata2pc for computing FPCA.

Examples

t <- seq(0, 1, length.out = 50)
X <- matrix(0, 30, 50)
for (i in 1:30) X[i, ] <- sin(2*pi*t + runif(1, 0, pi)) + rnorm(50, sd = 0.1)
fd <- fdata(X, argvals = t)
pc <- fdata2pc(fd, ncomp = 3)

# Plot PC components (mean +/- perturbations)
plot(pc, type = "components")


# Scree plot
plot(pc, type = "variance")


# Score plot
plot(pc, type = "scores")