spheres: Artificial data sets in three dimensions

In rwehrens/PCA: Functions for Principal Component Analysis

Artificial data sets in three dimensions

Description

The spheres, spheres2 and discs data sets are artificial data sets to illustrate dimension reduction by PCA and ICA. More details:

spheres:

1000 objects consisting of a combination of two sets of 500 samples each, drawn from two trivariate normal distributions with respective (empirical) means (10,10,0) and (25,25,0), and (empirical) covariance matrix diag(c(10,10,10)) in both cases. The entire set can be seen as two spheres with centres lying on the line y=x in the z=0 plane.

spheres2:

1000 objects consisting of a combination of two sets of 500 samples each, drawn from two trivariate normal distributions with respective (empirical) means (10,10,0) and (20,30,0), and (empirical) covariance matrix diag(c(10,10,10)) in both cases. The entire set can be seen as two spheres with centres lying on the line y=2x-10 in the z=0 plane.

discs:

1000 objects consisting of a combination of two sets of 500 samples each, drawn from two trivariate normal distributions with respective (empirical) means (40,36,44) and (40,44,36), and (empirical) covariance matrix matrix(c(30,0,21,0,30,21,21,21,30), 3, 3) in both cases. The entire set can be seen as two disc-like ellipsoids that lie above eachother along the same direction, but are slightly shifted with respect to eachother.

Usage

data(spheres)

rwehrens/PCA documentation built on Jan. 27, 2024, 2:33 p.m.