generate.toydata: Toy data generation

In dmt: Dependency Modeling Toolkit

Description

Generate simulated data which follows the distributional assumptions of the model.

Usage

generate.toydata(N = 100, zDim = 2, xDim = 3, yDim = 3, marginal.covariances = "full", priors = NULL)

Arguments

N	Sample size.
zDim	Dimensionality of the latent variable.
xDim	Dimensionality of X data set.
yDim	Dimensionality of Y data set.
marginal.covariances	"full": full covariance matrices for marginal noise (assumed by pCCA); "diagonal": diagonal covariances for marginal noise (pFA); "isotropic": isotropic covariances (pPCA).
priors	Set priors for toydata generation. Use as in fit.dependency.model.

Details

Assuming normally distributed latent variables for shared component Z, and data-specific components Zx, Zy. These follow standard multivariate normal distribution N(0, I). The observations X and Y are obtained as X = Wx*Z + Bx*Zx, Y = Wy*Z + By*Zy.

Value

List with the following components:

Z, Zx, Zy	Shared and data-set specific latent variables.
Wx, Wy, Bx, By	Transformation matrices.
X, Y	Data sets.

Author(s)

Leo Lahti [e-mail]

References

See citation("dmt") for references.

Examples

 toy <- generate.toydata(N = 100, zDim = 1, xDim = 3, yDim = 3, marginal.covariances = "full") 

dmt documentation built on May 2, 2019, 5:49 p.m.