pseudo: Pseudoinverse.

In FIACH: Retrospective Noise Control for fMRI

Description

This function computes the Pseudoinverse of a design matrix. If a matrix of dependent variables is also supplied the betas will also be returned. Optionally the residuals may be returned as well.

Usage

pseudo(x,y=NULL,residuals=FALSE,keepMean=FALSE,includeIntercept=TRUE)

Arguments

x	x must be a numeric matrix.
y	y must be a numeric matrix.
residuals	residuals must be either TRUE or FALSE.
keepMean	keepMean must be either TRUE or FALSE. This argument specifies whether residuals should be zero centered or retain their mean.
includeIntercept	includeIntercept must be either TRUE or FALSE. This argument specifies whether a column of ones should be included in x.

Value

Returns a Matrix containing either the pseudoinverse, linear regression coefficients or residuals.

Examples

data(trees, package="datasets")

## with intercept
pinv<-pseudo(x=log(trees$Girth),includeIntercept=TRUE)
## without intercept
pinv<-pseudo(x=log(trees$Girth),includeIntercept=FALSE)
##coefficients
coef<-pseudo(x=log(trees$Girth),y=log(trees$Volume))
## residuals
res<-pseudo(x=log(trees$Girth),y=log(trees$Volume),residuals=TRUE)

## standard model
mod<-lm(log(trees$Volume)~log(trees$Girth))
coef2<-coefficients(mod)
res2<-residuals(mod)
## equality

cbind(coef,coef2) ## same but order reversed
cbind(res,res2)   ## same

FIACH documentation built on May 1, 2019, 8:02 p.m.