plspm (version 0.2-2)

plsca: PLS-CA: Partial Least Squares Canonical Analysis

Description

Performs partial least squares canonical analysis for two blocks of data. Compared to PLSR2, the blocks of variables in PLS-CA play a symmetric role (i.e. there is neither predictors nor predictands)

Usage

plsca(X, Y, nc = NULL, scaled = TRUE)

Arguments

X

A numeric matrix or data frame (X-block).

Y

A numeric matrix or data frame (Y-block).

nc

The number of extracted PLS components (NULL by default)

scaled

A logical value indicating whether scaling data should be performed (TRUE by default).

Value

• An object of class "plsca", basically a list with the following elements:
• x.scores
  scores of the X-block (also referred to as T components).
• x.wgs
  weights of the X-block.
• x.loads
  loadings of the X-block.
• y.scores
  scores of the Y-block (also referred to as U components).
• y.wgs
  weights of the Y-block.
• y.loads
  loadings of the Y-block.
• cor.xt
  correlations between X and T.
• cor.yu
  correlations between Y and U.
• cor.tu
  correlations between T and U.
• cor.xu
  correlations between X and U.
• cor.yt
  correlations between Y and T.
• R2X
  explained variance of X by T.
• R2Y
  explained variance of Y by U.
• com.xu
  communality of X with U.
• com.yt
  communality of Y with T.

Details

Arguments X and Y must contain more than one variable. No missing data are allowed. When nc=NULL the number of components is determined by taking the minimum between the number of columns from X and Y. When scaled=TRUE the data is scaled to standardized values (mean=0, variance=1). Otherwise the data will only be centered (mean=0).

References

Tenenhaus, M. (1998) La Regression PLS. Theorie et Pratique. Editions TECHNIP, Paris.

See Also

print.plsca,plot.plsca

Examples

## example of PLSCA with the vehicles dataset
  data(vehicles)
  can <- plsca(vehicles[,1:12], vehicles[,13:16])
  can
  plot(can)