wefa {ImpuR}R Documentation

window evolving factor analysis

Description

Compute a progressive principal component analysis (or window evolving factor analysis) for a data matrix.

Usage

wefa(data, margin = 5, hetero.correct = 4:8)

Arguments

data a matrix
margin margin arount current row to form the current window
hetero.correct if TRUE, correct principal components by average of the components selected by hetero.correct (heteroscedasticity correction)

Details

the matrix is iteratively analyzed over a sliding window of size 2*margin+1 rows. At each time point, a principal component is performed over the "slice" and the principal components are retained. By default, an average of the principal components from 4 to 8 is subtracted from the others. This takes away the effect of heteroscedasticity when there are only one or two real components and maybe one or two artefacts.

Value

a matrix containing the principal components as columns

Author(s)

Christian Ritter

References

Ritter and Gilliard (2007), JSS

See Also

PeakPurity,Wefa


[Package ImpuR version 1.0 Index]