biVarimax: biVarimax

View source: R/helper.R

biVarimaxR Documentation

biVarimax

Description

rotate a matrix pair (U, V) to have clear structures. Extended from varimax of a single matrix.

Usage

biVarimax(U, V, lambda = 1, normalize = TRUE, eps = 1e-05)

Arguments

lambda

hyper-parameter of balancing U and V: obj(U, V) = lambda * obj_varimax(U) + obj_varimax(V). As lambda increases, more emphasis is put on U. Default is 1.

normalize

logical; if true, the rows of U, V are re-scaled to unit length before rotation, and scaled back afterwards. Default is TRUE.

eps

the tolerance for stopping: the relative change in the sum of singular values. Default is 1e-5.

V:

phenotype - mediator matrix; each row represents a phenotype, each column represents a mediator. The number of mediators should be no larger than that of phenotypes.

U:

genotype - mediator matrix estimator; each row represents a genotype, each column represents a mediator. The number of mediators should be no larger than that of genotypes.


ZijunGao/latentMediator documentation built on April 24, 2022, 12:12 a.m.