[R] QR factorization for aov

[peter dalgaard]

On Jun 11, 2013, at 20:26 , Mimi Celis wrote:

> I am looking at the aov function in R. I see that it uses a modified QR factorization routine dqrdc2 based on the Linpack routine dqrdc. Pivoting is done different than the original Linpack function.
> 
> My questions:
> 
>  *   Why is it necessary to modify the pivoting strategy? Something necessary for aov?
>  *   Will the LAPACK function dgeqp3 work well or would the pivoting have o be modified?
> 


As far as I have understood it, the issue is that to generate the sequential ANOVA table based on a single QR factorization, you can't have it swapping terms around, because there are cases where the order of terms matters. 

So LAPACK won't do. Or rather, to use it, you'd need to rethink the algoritm; you may need refactorizing for each line the ANOVA table. It's not necessarily impossible or seriously inefficient, just one of the things that nobody seems to have an aching desire to tackle.

-pd

> Thank you
> MRC
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

-- 
Peter Dalgaard, Professor
Center for Statistics, Copenhagen Business School
Solbjerg Plads 3, 2000 Frederiksberg, Denmark
Phone: (+45)38153501
Email: pd.mes at cbs.dk  Priv: PDalgd at gmail.com