Re: [R] singular factor analysis

From: Patrick Burns <>
Date: Thu 07 Sep 2006 - 09:05:39 GMT

This is a very common computation in finance.

On the public domain page of the Burns Statistics website in the financial part is the code and R help file for 'factor.model.stat'. Most of the complication of the code is to deal with missing values.

Patrick Burns
+44 (0)20 8525 0696
(home of S Poetry and "A Guide for the Unwilling S User")

Spencer Graves wrote:

> Are there any functions available to do a factor analysis with
>fewer observations than variables? As long as you have more than 3
>observations, my computations suggest you have enough data to estimate a
>factor analysis covariance matrix, even though the sample covariance
>matrix is singular. I tried the naive thing and got an error:
> > set.seed(1)
> > X <- array(rnorm(50), dim=c(5, 10))
> > factanal(X, factors=1)
>Error in solve.default(cv) : system is computationally singular:
>reciprocal condition number = 4.8982e-018
> I can write a likelihood for a multivariate normal and solve it,
>but I wondered if there is anything else available that could do this?
> Thanks,
> Spencer Graves
> mailing list
>PLEASE do read the posting guide
>and provide commented, minimal, self-contained, reproducible code.
> mailing list PLEASE do read the posting guide and provide commented, minimal, self-contained, reproducible code. Received on Thu Sep 07 19:11:36 2006

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.1.8, at Thu 07 Sep 2006 - 17:46:09 GMT.

Mailing list information is available at Please read the posting guide before posting to the list.