tgpca: tgpca

In amaendle/mvTargetOpt: Multivariate multi-target optimization

Description

Function which performs a principal component analysis (PCA) on the descriptor variable data (in the target space) given by dat, In order to choose a certain direction through the target point for the projections, wg has to be set to 1 – then the target point is chosen as center for the PCA. If wg lies between 0 and 1, pseudo observations at the target point are created such that a ratio of wg of the observations are pseudo observations. Then prcomp is applied to the standardized data and pseudo data.

Usage

tgpca(
  dat,
  tgmean = NULL,
  tgerr = NULL,
  wg = 1,
  wfun,
  mknormweights,
  yweights = F,
  ylast = NULL
)

Arguments

dat	matrix, data.frame
tgmean	numeric vector, optional
tgerr	numeric vector, optional
wg	numeric, weight for the target value. If wg equals 1 or 2 then the pca is performed with the target value as center
wfun	function, weight function
mknormweights	unused
yweights	boolean, use weights?
ylast	integer or NULL, if integer, ignore observations older than the last ylast evaluation points.

Value

returns the results of the pca and some extra stuff

Examples

tgpca(matrix(rnorm(20),ncol=2), tgmean=c(0,0))

amaendle/mvTargetOpt documentation built on June 12, 2020, 5:57 p.m.