[R] Design matrix not identity

[Prof Brian Ripley]

On Wed, 12 May 2004, Patrik Waldmann wrote:

> I was too quick before. What I was looking for was a function that
> constructs the design (or incidence) matrix (X in a linear model) from a
> factor. Uwe Ligges suggested using model.matrix and this does almost what I
> want, but it is first necessary to construct a data variable. 

Eh?  You have to construct the factor, and nothing else.

> It also asigns ones to all rows of the first column (because this is set
> to be the contrast, not really what I want - see below). Maybe time for
> a function that just converts a factor into a design matrix?

Uwe was quite correct, and you were still too quick.

[Don't call an object after an R function. Let's use a sensible name like
`f'.]

f <- as.factor(c(1,1,2,2,3,3,3))
model.matrix(~ 0 + f)

or

diag(nlevels(f))[f,]

gives what you illustrate.

> I have a factor
> factor<-as.factor(c(1,1,2,2,3,3,3))
> 
> and I want a matrix
> 1 0 0
> 1 0 0
> 0 1 0
> 0 1 0
> 0 0 1
> 0 0 1
> 0 0 1

> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html

PLEASE do.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595