[18692] in s-news-athena
[S] summary and more questions on help of "control" of lme function
daemon@ATHENA.MIT.EDU (sjcav@aros.net)
Wed Jul 21 16:09:57 1999
Message-Id: <3.0.1.32.19990721140330.0076546c@mail.aros.net>
Date: Wed, 21 Jul 1999 14:03:30 -0600
To: s-news@wubios.wustl.edu
From: sjcav@aros.net
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Thanks to Joel Dubin, Frank Ashford and Douglas Bates for their replies.
Expression 1:
lme(fixed = Dif ~ DifLag1+DifLag2, cluster = ~Contract,
data = CCDeliveryPeriod, control =
list(maxiter=100,max.maxiter=100,ms.maxiter = 100,
verbose = T))
is an accepted expression for SPlus
Expression 2:
lme(fixed = Dif ~ DifLag1+DifLag2, cluster = ~Contract,
data = CCDeliveryPeriod,
control=lme.control(maxiter=100,max.maxiter=100,ms.maxiter = 100,
verbose = T))
I got the "dumped" message again.
I am using Splus4.3 for windows.
However, for expression 1, I got the warning message:
Warning messages:
ITERATION LIMIT REACHED WITHOUT OTHER CONVERGENCE in: ms( ~ lme.loglik.i
dentity(theta, a, q, p, M), start = list(theta = a$ ...
When I use expression (1) and the summary function to see the regression
results, I got
Estimation Method: RML
Convergence at iteration: 50
Restricted Loglikelihood: 6190.893049
Restricted AIC: -12309.7861
Restricted BIC: -12064.69557
The "convergence at iteration" is 50 (the default), instead what I set in
control "100". Is my understanding right? Does the control work in this
case? When I set
control = list(maxiter=20,max.maxiter=20,ms.maxiter = 20,
verbose = T)),
The regression results are the same as Expression (1), and the convergence
at iteration still says 50.
I would appreciate your help.
---Sharon Cavenaugh
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
