[R-sig-ME] glmutli package assistance please

Discussion:

Bill Poling

2018-11-15 11:46:48 UTC

Hi, I have removed the pdf which was causing my e-mail to be blocked by moderators, my apologies.

https://www.jstatsoft.org/article/view/v034i12/v34i12.pdf

Original post:

Hello. I am still trying to get some of the examples in this glmulti pdf to work with my data.

I have sent e-mails to author addresses provided but no response or bounced back as in valid.

I am not sure if this is more likely to receive support on r-help or r-sig-mixed-models, hence the double posting, my apologies in advance.

I am windows 10 -- R3.5.1 -- RStudio Version 1.1.456

glmulti: An R Package for Easy Automated Model Selection with (Generalized) Linear Models

pdf Attached:

On page 13 section 3.1 of the pdf they describe a routine to estimate the candidate models possible.

Their data description:
The number of levels factors have does not affect the number of candidate models, only their complexity. We use a data frame dod, containing as a first column a dummy response variable, the next 6 columns are dummy factors with three levels, and the last six are dummy covariates.
To compute the number of candidate models when there are between 1 and 6 factors and 1 and 6 covariates, we call glmulti with method = "d" and data = dod. We use names(dod) to specify the names of the response variable and of the predictors. We vary the number of factors and covariates, this way:

Their routine:
dd <- matrix(nc = 6, nr = 6) for(i in 1:6) for(j in 1:6) dd[i, j] <- glmulti(names(dod)[1],
+ names(dod)[c(2:(1 + i), 8:(7 + j))], data = dod, method = "d")

My data, I organized it similar to the example, Response, Factor, Factor, 5 covariates

Classes 'data.table' and 'data.frame':23141 obs. of 8 variables:
$ Editnumber2 : num 0 0 1 1 1 1 1 1 1 1 ...
$ PatientGender : Factor w/ 3 levels "F","M","U": 1 1 2 2 2 2 1 1 1 1 ...
$ B1 : Factor w/ 14 levels "Z","A","C","D",..: 2 2 3 3 2 2 2 2 2 2 ...
$ SavingsReversed: num -0.139 -0.139 -0.139 -0.139 -0.139 ...
$ productID : int 3 3 3 3 3 3 3 3 1 1 ...
$ ProviderID : int 113676 113676 113964 113964 114278 114278 114278 114278 114278 114278 ...
$ ModCnt : int 0 0 0 0 1 1 1 1 1 1 ...
$ B2 : num -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 ...
- attr(*, ".internal.selfref")=<externalptr>

Trying to follow what they did, my routine, Editnumber2 is the response variable:

dd <- matrix(nc = 2, nr = 5)
for(i in 1:2) for(j in 1:5) dd[i, j] <- glmulti(names(r1)[1], names(r1)[c(2:(1 + i), 7:(6 + j))], data = r1, method = "d")

The error: Error in terms.formula(formula, data = data) :
invalid model formula in ExtractVars

I have tried changing the numbers around but get results like this:

Initialization...
TASK: Diagnostic of candidate set.
Sample size: 23141
2 factor(s).
2 covariate(s). <--appears to be missing 3 of the covariates for some reason?
0 f exclusion(s).
0 c exclusion(s).
0 f:f exclusion(s).
0 c:c exclusion(s).
0 f:c exclusion(s).
Size constraints: min = 0 max = -1
Complexity constraints: min = 0 max = -1 Your candidate set contains 250 models.
Error in `[<-`(`*tmp*`, i, j, value = glmulti(names(r1)[1], names(r1)[c(2:(1 + :
subscript out of bounds

I hope someone can help straighten out my code, thank you.

WHP

Confidentiality Notice This message is sent from Zelis. ...{{dropped:13}}

Bert Gunter

2018-11-15 15:43:00 UTC

Permalink

Please do not cross post (see te posting guide). This should go only
to the mixed models list.

-- Bert

Bert Gunter

"The trouble with having an open mind is that people keep coming along
and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )

Post by Bill Poling
Hi, I have removed the pdf which was causing my e-mail to be blocked by moderators, my apologies.
https://www.jstatsoft.org/article/view/v034i12/v34i12.pdf
Hello. I am still trying to get some of the examples in this glmulti pdf to work with my data.
I have sent e-mails to author addresses provided but no response or bounced back as in valid.
I am not sure if this is more likely to receive support on r-help or r-sig-mixed-models, hence the double posting, my apologies in advance.
I am windows 10 -- R3.5.1 -- RStudio Version 1.1.456
glmulti: An R Package for Easy Automated Model Selection with (Generalized) Linear Models
On page 13 section 3.1 of the pdf they describe a routine to estimate the candidate models possible.
The number of levels factors have does not affect the number of candidate models, only their complexity. We use a data frame dod, containing as a first column a dummy response variable, the next 6 columns are dummy factors with three levels, and the last six are dummy covariates.
dd <- matrix(nc = 6, nr = 6) for(i in 1:6) for(j in 1:6) dd[i, j] <- glmulti(names(dod)[1],
+ names(dod)[c(2:(1 + i), 8:(7 + j))], data = dod, method = "d")
My data, I organized it similar to the example, Response, Factor, Factor, 5 covariates
$ Editnumber2 : num 0 0 1 1 1 1 1 1 1 1 ...
$ PatientGender : Factor w/ 3 levels "F","M","U": 1 1 2 2 2 2 1 1 1 1 ...
$ B1 : Factor w/ 14 levels "Z","A","C","D",..: 2 2 3 3 2 2 2 2 2 2 ...
$ SavingsReversed: num -0.139 -0.139 -0.139 -0.139 -0.139 ...
$ productID : int 3 3 3 3 3 3 3 3 1 1 ...
$ ProviderID : int 113676 113676 113964 113964 114278 114278 114278 114278 114278 114278 ...
$ ModCnt : int 0 0 0 0 1 1 1 1 1 1 ...
$ B2 : num -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 ...
- attr(*, ".internal.selfref")=<externalptr>
dd <- matrix(nc = 2, nr = 5)
for(i in 1:2) for(j in 1:5) dd[i, j] <- glmulti(names(r1)[1], names(r1)[c(2:(1 + i), 7:(6 + j))], data = r1, method = "d")
invalid model formula in ExtractVars
Initialization...
TASK: Diagnostic of candidate set.
Sample size: 23141
2 factor(s).
2 covariate(s). <--appears to be missing 3 of the covariates for some reason?
0 f exclusion(s).
0 c exclusion(s).
0 f:f exclusion(s).
0 c:c exclusion(s).
0 f:c exclusion(s).
Size constraints: min = 0 max = -1
Complexity constraints: min = 0 max = -1 Your candidate set contains 250 models.
subscript out of bounds
I hope someone can help straighten out my code, thank you.
WHP
Confidentiality Notice This message is sent from Zelis. ...{{dropped:13}}
______________________________________________
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Michael Dewey

2018-11-15 15:52:48 UTC

Permalink

Dear Bert

Since glmulti operates on glm/lm models I think, although I agree about
not cross-posting, that it was OK here. Perhaps I do not understand the
full significance of mixed models though.

Michael

Post by Bert Gunter
Please do not cross post (see te posting guide). This should go only
to the mixed models list.
-- Bert
Bert Gunter
"The trouble with having an open mind is that people keep coming along
and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )

______________________________________________
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

--
Michael
http://www.dewey.myzen.co.uk/home.html

Bert Gunter

2018-11-15 16:03:09 UTC

Permalink

OK. Then post here but *not* on mixed models list. One or the other,
exclusive or.

-- Bert
Bert Gunter

"The trouble with having an open mind is that people keep coming along
and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )

Post by Michael Dewey
Dear Bert
Since glmulti operates on glm/lm models I think, although I agree about
not cross-posting, that it was OK here. Perhaps I do not understand the
full significance of mixed models though.
Michael

--
Michael
http://www.dewey.myzen.co.uk/home.html

Bill Poling

2018-11-15 18:00:18 UTC

Permalink

Hello Michael, thank you for your responses.

I will have to look through the myriad of iterations of attempts at all this on this project but it seems to me that I tried that? Or tried to convert the integers to numeric and it errored out?

If not I certainly will do so and let know, thank you Sir!

WHP

From: Michael Dewey <***@dewey.myzen.co.uk>
Sent: Thursday, November 15, 2018 7:24 AM
To: Bill Poling <***@zelis.com>; r-sig-mixed-***@r-project.org; r-help (r-***@r-project.org) <r-***@r-project.org>
Subject: Re: [R] glmutli package assistance please

Dear Bill

I am not sure what is going on here but I notice that 2 of your
covariates are numeric and 3 integer. What happens if you make them all
numeric?

Michael

--
Michael
http://www.dewey.myzen.co.uk/home.html

Confidentiality Notice This message is sent from Zelis. This transmission may contain information which is privileged and confidential and is intended for the personal and confidential use of the named recipient only. Such information may be protected by applicable State and Federal laws from this disclosure or unauthorized use. If the reader of this message is not the intended recipient, or the employee or agent responsible for delivering the message to the intended recipient, you are hereby notified that any disclosure, review, discussion, copying, or taking any action in reliance on the contents of this transmission is strictly prohibited. If you have received this transmission in error, please contact the sender immediately. Zelis, 2018.