canonical_correlation_analysis: Canonical correlation analysis

In jandob/ccf: Canonical correlation forest

Description

Canonical correlation analysis (CCA) finds pairs of vectors (w,v) such that projections Xw and Yv have maximal possible correlations. The pairs are ordered in decreasing order of the correlations. In addition, projection vectors are normalized such that the variance of Xw and of Yv is equal to 1. This means that projections are not only correlated, but "on the same scale" and hence can be directly compared.

Usage

canonical_correlation_analysis(x, y, epsilon = 1e-04)

Arguments

x	Matrix of size n-by-p with n observations from p variables. Alternatively, data frames and numeric vectors are supported and automatically converted.
y	Matrix of size n-by-p with n observations from p variables. Alternatively, data frames and numeric vectors are supported and automatically converted.
epsilon	Numeric value usued as tolerance threshold for rank reduction of the input matrices. Default is 1e-4.

Value

A list containing the following components

• xcoefEstimated estimated coefficients for the x variable.

• ycoefEstimated estimated coefficients for the y variable.

• corMatrix with correlation coefficients.

Examples

library(MASS)
library(pracma)

X <- mvrnorm(1000, mu = c(0, 0), Sigma = eye(2))
cca <- canonical_correlation_analysis(X, X)
cca

X <- mvrnorm(1000, mu = c(1, 2),
             Sigma = matrix(c(1.5, 0.5, 0.5, 1.5), ncol = 2))
cca <- canonical_correlation_analysis(X, X)
cca

jandob/ccf documentation built on May 18, 2019, 12:23 p.m.