s-news
[Top] [All Lists]

Warnings in censorReg

To: s-news@lists.biostat.wustl.edu
Subject: Warnings in censorReg
From: Karl "Hülber" <huelber@yahoo.com>
Date: Tue, 25 Jan 2005 09:08:47 -0800 (PST)
Comment: DomainKeys? See http://antispam.yahoo.com/domainkeys
Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; b=gHzxhHnIq2qV+dNYNMiaiZBpV2KxvkZc4aEBgmDOfwxO3hB6d0G9kV5vGOCOpQ2RUKJ/mkeOhBpxWQOznMtGRJNUSxobPpuWwRDbWbMlFLmzoS9YGfvJs7dwAnz6pA4uYjb2dFeBaSs6TejH2RA3wEBE32NyJ6mOhuanaqwXI9c= ;
In-reply-to: <20050112163420.310444026C7@smtp.biostat.wustl.edu>
Hi!

I´m trying to fit a parametric survival model for
interval censored data using censorReg in S-Plus 2000.

Unfortunately I faced the following error:
Error in length(p) <- m: Cannot allocate 187196208
bytes: options("object.size") is 5000000: see options
help file

Changing the object.size-parameter could not save the
problem, because this prolongs computing times to
infinity. The error is caused ? as far as I found out
? by ties in the data. Can anybody tell me how to deal
with ties in interval censored data?

Additionally, these warnings occurred:
1) Warning messages: Warning in censorReg.mlest(Yi,
censor.codesi, Xi (or NULL), casewti, distribution,
link, : Information matrix is not positive definite
2) Warning messages: Warning in kaplanMeier.fit(y,
censor.codes, case.weights): Cdfest error message 8

How strong is their impact on the estimation of
coefficients and is there a way to avoid the warnings?

Any help would be greatly appreciated
Karl Hülber



                
__________________________________ 
Do you Yahoo!? 
Yahoo! Mail - Easier than ever with enhanced search. Learn more.
http://info.mail.yahoo.com/mail_250

<Prev in Thread] Current Thread [Next in Thread>
  • Warnings in censorReg, Hülber <=