[R] factanal frustration!
Prof Brian Ripley
ripley at stats.ox.ac.uk
Mon Jul 9 22:32:20 CEST 2007
factanal() does not have an argument method="mle".
Trying to do factor analysis on 1000 columns is quite unrealistic, but you
may find rescaling the matrix helps.
On Mon, 9 Jul 2007, Talbot Katz wrote:
> It seems that nearly every time I try to use factanal I get the following
> Error in factanal(mretdb1, factors = 2, method = "mle", control =
> list(nstart = 25)) :
> unable to optimize from these starting value(s)
> In the case cited above, mretdb1 is synthetic data created from two factors,
> that is, mretdb1 has dimensions 3050*1000, and each of the thousand columns
> has the form
> c[i]=a1[i]*f1 + a2[i]*f2 + a3[i]*fn[,i]
> where a1, a2, a3 are each vectors of length 1000, f1 and f2 are vectors of
> length 3050, and fn is a matrix of dimension 3050*1000, such that f1, f2,
> and all the columns of fn are mutually uncorrelated. I wanted to see how
> closely the factors returned by factanal are to the ones that generated the
> data. But, as you can see, factanal hasn't cooperated!
> I would greatly appreciate advice from factanal users, especially those who
> have experienced the same error. Thanks!
> -- TMK --
> 917-656-5351 cell
> 212-460-5430 home
> R-help at stat.math.ethz.ch mailing list
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
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
More information about the R-help