MBplotBlocks: Customizable plots of blocks related information

In MBAnalysis: Multiblock Exploratory and Predictive Data Analysis

Customizable plots of blocks related information

Description

Plots blocks related information of MBPCA, ComDim, MBPLS or MBWCov with several options of customization.

Usage

MBplotBlocks(
  res,
  which = "explained.blocks&Y",
  axes = c(1, 2),
  blocks.axes = 1:max(axes),
  title = NULL,
  size = 2.25
)

Arguments

res	An object resulting from MBPCA, ComDim, MBPLS or MBWCov.
which	Either "explained.blocks&Y", "scree", "structure" or "blocks.axes". See details.
axes	Which global dimensions should be plotted? Only useful if which=structure or which=blocks.axes
blocks.axes	Which individual blocks dimensions should be correlated with global ones? Only useful if which=blocks.axes
title	An optional title to be added to the plot.
size	The overall size of labels, points, etc.

Details

• explained.blocks&Y: Barplot of the percentages of inertia explained in each block of variables (and Y for MBPLS or MBWCov) by each global components.

• scree: Barplot of the saliences of each block of variables on each global components.

• structure: Blocks coordinates (saliences) on the global selected axes

• blocks.axes: Correlations of the selected individual blocks.axes with the global selected axes.

Value

The required plot.

See Also

plot.MBPCA plot.ComDim plot.MBPLS plot.MBWCov

Examples

# Unsupervised example

data(ham)
X=ham$X
block=ham$block
res.mbpca <- MBPCA(X,block, name.block=names(block))
MBplotBlocks(res.mbpca,which="explained.blocks&Y")
MBplotBlocks(res.mbpca,which="scree")
MBplotBlocks(res.mbpca,which="structure")
MBplotBlocks(res.mbpca,which="blocks.axes")

# Supervised example

data(ham)
X=ham$X
block=ham$block
Y=ham$Y
res.mbpls <- MBPLS(X, Y, block, name.block=names(block))
MBplotBlocks(res.mbpls,which="explained.blocks&Y")
MBplotBlocks(res.mbpls,which="scree")
MBplotBlocks(res.mbpls,which="structure")
MBplotBlocks(res.mbpls,which="blocks.axes")

MBAnalysis documentation built on Oct. 24, 2023, 5:07 p.m.