Seeded Canonical Correlation Analysis
Coefficients of ordinary and partial least squares through iterative p...
scree-ploting cov(X, Y)
finalized CCA in seeded CCA
Fitted values of ordinary and partial least squares
Initialized CCA in seeded CCA
Plotting class "seedCCA" depending on the value of type
Projection of a seed matrix on to the column subspace of M with respec...
basic function for printing class "seedCCA"
Seeded Canonical correlation analysis
increments of iterative projections with automatic stopping
increments of iterative projections
Ordinary least squares
Partial least squares through iterative projections
Function that guides a selection of the terminating index when using s...
Functions for dimension reduction through the seeded canonical correlation analysis are provided. A classical canonical correlation analysis (CCA) is one of useful statistical methods in multivariate data analysis, but it is limited in use due to the matrix inversion for large p small n data. To overcome this, a seeded CCA has been proposed in Im, Gang and Yoo (2015) \doi{10.1002/cem.2691}. The seeded CCA is a two-step procedure. The sets of variables are initially reduced by successively projecting cov(X,Y) or cov(Y,X) onto cov(X) and cov(Y), respectively, without loss of information on canonical correlation analysis, following Cook, Li and Chiaromonte (2007) \doi{10.1093/biomet/asm038} and Lee and Yoo (2014) \doi{10.1111/anzs.12057}. Then, the canonical correlation is finalized with the initially-reduced two sets of variables.