[6505] in s-news-athena
missing values ?
daemon@ATHENA.MIT.EDU (Byron Bodo)
Fri Feb 24 08:35:21 1995
Date: Fri, 24 Feb 1995 08:00:35 -0500
To: s-news@utstat.toronto.edu
From: bodo@io.org (Byron Bodo)
Handling missive values seems to be an ugly dirty business
in S-plus. I'm working with large multivariate data matrices that are
typically riddled with missing values. Most S-plus (Win 3.1 vers)
won't accept the data, and there's precious little information
in the manuals or other material to suggest work arounds.
I've fumbled about with *apply* functions, but still don't
have a sense of the best way to handle missing values.
To pose a simple problem, say it's required to derive rank correlation
matrices from an n x m matrix containing missing values by
(a) listwise, and (b) pairwise reduction. Is there any
simple way to do this in S-plus, or is it easier to write a FORTRAN
routine?
-bb
Byron Bodo
240 Markham St., Toronto, ON, Canada M6J 2G6
tel: (416) 967 7286
fax: (416) 967 9004
email: bodo@io.org
