STATegRa (version 1.0.0)

PCA.selection: Selection of optimal number of components using PCA

Description

Function to select the number of optimal components using Principal Components Analysis. There are four different criterions to select the number of components: accumulated variance explained, the individual explained variance of each component, the absolute value of its variability or just a fixed number of components.

Usage

PCA.selection(Data,fac.sel,varthreshold=NULL,nvar=NULL,PCnum=NULL)

Arguments

Data

Matrix with data. Samples has to be in columns and variables in rows

fac.sel

Criterium for selecting number of components. The posible option are: "%accum", "single%", "rel.abs" and "fixed.num"

varthreshold

Threshold for the selection of components in "%accum", "single%" criterions

nvar

Threshold applied when the option "rel.abs" is selected

PCnum

Fixed number of components to select when the option"fixed.num" is selected

Value

The function returns a list with the following components:

PCAres

A list containing the results of the PCA decomposition

numComps

Number of selected components applying the selected criterion

The components of PCAres list are the following:

eigen

Eigen values of the decomposition

var.exp

Explained variance of each component

scores

Matrix of scores

loadings

Matrix of loadings

See Also

modelSelection

Examples

data(STATegRa_S3)
ps <- PCA.selection(Data=Block2.PCA,fac.sel="single%",varthreshold=0.03)
ps$numComps