Hello Ben:
Thanks for replying.
Unfortunately, it's hard for me or anyone else to determine when the copyright is violated. For instance, "The elements of statistical learning" is available on Amazon for a hefty price of $63, and yet the authors made the pdf freely available as well. I assumed that the same was true for Searle's book.
I have developed an implementation of Searle's algorithm based on the equations 96-99 and 91b. On top of that, it uses the multiplication trick to incorporate weights.It generates results that are close to those of PROC MIXED when there are no weights, or when weights are used w/o
random effects. However, when weights are used with random effects, the results are off. Did I get it right that, in your opinion, I should look for some error in the code?
Regards,Nik
From: "r-sig-mixed-models-***@r-project.org" <r-sig-mixed-models-***@r-project.org>
To: r-sig-mixed-***@r-project.org
Sent: Tuesday, October 16, 2018 7:27 PM
Subject: R-sig-mixed-models Digest, Vol 142, Issue 15
Send R-sig-mixed-models mailing list submissions to
r-sig-mixed-***@r-project.org
To subscribe or unsubscribe via the World Wide Web, visit
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
or, via email, send a message with subject or body 'help' to
r-sig-mixed-models-***@r-project.org
You can reach the person managing the list at
r-sig-mixed-models-***@r-project.org
When replying, please edit your Subject line so it is more specific
than "Re: Contents of R-sig-mixed-models digest..."
Today's Topics:
1. Dependency structure (Yashree Mehta)
2. Weighted regression in REML (Nik Tuzov)
3. Weighted regression in REML (Nik Tuzov)
4. Weighted regression in REML (Nik Tuzov)
5. Weighted regression in REML (Nik Tuzov)
6. Re: Weighted regression in REML (Ben Bolker)
----------------------------------------------------------------------
Message: 1
Date: Tue, 16 Oct 2018 19:03:27 +0200
From: Yashree Mehta <***@gmail.com>
To: r-sig-mixed-***@r-project.org
Subject: [R-sig-ME] Dependency structure
Message-ID:
<CAOE=hqLskrB+f1fnq2Uw-XYNbVyenYby4o=_ZecK5336D-=***@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"
Hi,
Is there literature on how to specify the dependency structure between the
random intercept and the statistical noise error term in a random intercept
model?
It would be useful to also know how to implement using R...
Thank you
Yashree
[[alternative HTML version deleted]]
------------------------------
Message: 2
Date: Tue, 16 Oct 2018 21:14:56 +0000 (UTC)
From: Nik Tuzov <***@ntuzov.com>
To: "r-sig-mixed-***@r-project.org"
<r-sig-mixed-***@r-project.org>
Subject: [R-sig-ME] Weighted regression in REML
Message-ID: <***@mail.yahoo.com>
Content-Type: text/plain; charset="utf-8"
Hello:
I'm reading the book of Searle:
http://www.leg.ufpr.br/~eder/Variance%20Components.pdf
I think that equation 53 on p 276 suggests that if one wants to use weights in REML, the solution is very similar to that in OLS: multiply X, Z, and Y by the square root of the weight matrixand proceed as in unweighted case. Am I right or there is more that needs to be done?
Regards,Nik
[[alternative HTML version deleted]]
------------------------------
Message: 3
Date: Tue, 16 Oct 2018 21:17:39 +0000 (UTC)
From: Nik Tuzov <***@ntuzov.com>
To: "r-sig-mixed-***@r-project.org"
<r-sig-mixed-***@r-project.org>
Subject: [R-sig-ME] Weighted regression in REML
Message-ID: <***@mail.yahoo.com>
Content-Type: text/plain; charset="utf-8"
Hello:
I'm reading the book of Searle:
http://www.leg.ufpr.br/~eder/Variance%20Components.pdf
I think that equation 53 on p 276 suggests that if one wants to use weights in REML, the solution is very similar to that in OLS: multiply X, Z, and Y by the square root of the weight matrixand proceed as in unweighted case. Am I right or there is more that needs to be done?
Regards,Nik
[[alternative HTML version deleted]]
------------------------------
Message: 4
Date: Tue, 16 Oct 2018 21:16:07 +0000 (UTC)
From: Nik Tuzov <***@ntuzov.com>
To: "r-sig-mixed-***@r-project.org"
<r-sig-mixed-***@r-project.org>
Subject: [R-sig-ME] Weighted regression in REML
Message-ID: <***@mail.yahoo.com>
Content-Type: text/plain; charset="utf-8"
Hello:
I'm reading the book of Searle:
http://www.leg.ufpr.br/~eder/Variance%20Components.pdf
I think that equation 53 on p 276 suggests that if one wants to use weights in REML, the solution is very similar to that in OLS: multiply X, Z, and Y by the square root of the weight matrixand proceed as in unweighted case. Am I right or there is more that needs to be done?
Regards,Nik
[[alternative HTML version deleted]]
------------------------------
Message: 5
Date: Tue, 16 Oct 2018 21:21:48 +0000 (UTC)
From: Nik Tuzov <***@ntuzov.com>
To: "r-sig-mixed-***@r-project.org"
<r-sig-mixed-***@r-project.org>
Subject: [R-sig-ME] Weighted regression in REML
Message-ID: <***@mail.yahoo.com>
Content-Type: text/plain; charset="utf-8"
Hello:
I'mreading the book of Searle:
http://www.leg.ufpr.br/~eder/Variance%20Components.pdf
Ithink that equation 53 on p 276 suggests that if one wants to useweights in REML, the solution is very similar to that in OLS:multiply X, Z, and Y by the square root of the weight matrix andproceed as in unweighted case. Am I right or there is more that needsto be done?
Regards, Nik
[[alternative HTML version deleted]]
------------------------------
Message: 6
Date: Tue, 16 Oct 2018 20:27:36 -0400
From: Ben Bolker <***@gmail.com>
To: r-sig-mixed-***@r-project.org
Subject: Re: [R-sig-ME] Weighted regression in REML
Message-ID: <a483cb3b-b15a-d30f-b7b2-***@gmail.com>
Content-Type: text/plain; charset="utf-8"
Can I please request that in general people **not** use this list to
post links to material that is (presumably) violating the
authors'/publisher's wishes? (A screenshot of a particular equation
would seem to constitute "fair use" ...)
FWIW your statement sounds correct to me -- in the Bates et al.
JSS/lmer paper (available via vignette("lmer", package="lme4")), the
development of the estimation procedure mostly leaves the weights out
for simplicity, but says:
"To allow for case weights, we save the products X^⊤ W X, X^⊤ W y, Z^⊤ W
X, Z^⊤ W y and Z^⊤ W Z (see Table 6)."
W is the weight matrix, so this is equivalent to multiplying X, y, Z
by W^(1/2) ...
cheers
Ben Bolker
Post by Nik Tuzovhttp://www.leg.ufpr.br/~eder/Variance%20Components.pdf
I think that equation 53 on p 276 suggests that if one wants to use weights in REML, the solution is very similar to that in OLS: multiply X, Z, and Y by the square root of the weight matrixand proceed as in unweighted case. Am I right or there is more that needs to be done?
Regards,Nik
[[alternative HTML version deleted]]
_______________________________________________
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
------------------------------
Subject: Digest Footer
_______________________________________________
R-sig-mixed-models mailing list
R-sig-mixed-***@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
------------------------------
End of R-sig-mixed-models Digest, Vol 142, Issue 15
***************************************************
[[alternative HTML version deleted]]