greedymatch: Greedy Match

In singR: Simultaneous Non-Gaussian Component Analysis

Greedy Match

Description

Greedy Match matches a column of Mx and My by minimizing chordal distance between vectors, removes the matched columns and then finds the next pair. This equivalent to maximizing absolute correlation for data in which each column has mean equal to zero. Returns permuted columns of Mx and My. This function does not do any scaling or sign flipping. For this matching to coincide with angle matching, the columns must have zero mean.

Usage

greedymatch(Mx, My, Ux, Uy)

Arguments

Mx	Subject Score for X with n x n.comp.X matrix
My	Subject Score for Y with n x n.comp.Y matrix
Ux	Matrix with n.comp x n, Mx = Lx^-1 %*% t Ux, Lx is the whitener matrix of dX.
Uy	Matrix with n.comp x n, My = Ly^-1 %*% t Uy, Ly is the whitener matrix of dY.

Value

a list of matrices:

Mx

Columns of original Mx reordered from highest to lowest correlation with matched component in My

My

Columns of original My reordered from highest to lowest correlation with matched component in Mx

Ux

Permuted rows of original Ux corresponds to MapX

Uy

Permuted rows of original Uy corresponds to MapY

correlations

a vector of correlations for each pair of columns in permuted Mx and M

mapX

the sequence of the columns in original Mx.

mapY

the sequence of the columns in original MY.

singR documentation built on May 29, 2024, 7:30 a.m.