[R] Package norm has been removed. What to use for Maximum Likelihood Missing Data Imputation?
Daniel Abbott
I apologize if an answer to my questions is available, or if I
submitted this question incorrectly. I have read the mailing lists, as
well as the R Project and CRAN homepages. However, I may have missed
something.
I noticed the package 'norm' has been removed. Its page
http://cran.r-project.org/web/packages/norm/index.html now reads:
"Package ‘norm’ was removed from the CRAN repository.
Formerly available versions can be obtained from the archive."
My questions are:
1. Why was norm removed? I used the package and found it to be very
useful. Is there a serious error in the package or another problem?
2. If norm should no longer be removed, what is another good package
for ML (maximum likelihood) missing data imputation? I have seen
several recommendations online, but I wonder which one is currently
the "reigning champ."
Thank you very much. I appreciate your time!
______________________________________________
R-h...@r-project.org mailing list
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.
Uwe Ligges
the checks for recent versions of R anymore.
If you think the package is very useful for you, you may want to become
its new active maintainer. Just take the sources from the archive, fix
and update the package as you like and submit a version that passes the
checks to CRAN.
Best,
Uwe Ligges
Daniel Abbott
I am mostly interested with ML missing data imputation. Is there
another R module which has such a function that is well regarded?
Thank you.
Achim Zeileis
> Hello,
>
> I apologize if an answer to my questions is available, or if I
> submitted this question incorrectly. I have read the mailing lists, as
> well as the R Project and CRAN homepages. However, I may have missed
> something.
>
> I noticed the package 'norm' has been removed. Its page
> http://cran.r-project.org/web/packages/norm/index.html now reads:
>
> "Package ?norm? was removed from the CRAN repository.
>
> Formerly available versions can be obtained from the archive."
>
> My questions are:
>
> 1. Why was norm removed? I used the package and found it to be very
> useful. Is there a serious error in the package or another problem?
What do you consider to be a serious error?
Typical reasons for removing a package from the active CRAN repository are
that the package does not pass CRAN checks and/or that the package
maintainer is irresponsive. (norm, specifically, hadn't been updated since
2002.) This is serious enough to make all automatic CRAN features such as
daily package checks and building of binary packages too cumbersome.
But if you do not consider this to be serious, you can keep on using
"norm", it is still in CRAN's package archives. Or, even better, you could
adopt it, fix it, and release a new version to CRAN (although it is not
quite clear to me whether the norm's license allows this).
(Note that CRAN runs no checks whether the methods implemented are
reasonable or well-suited for the problem they are trying to address etc.)
> 2. If norm should no longer be removed, what is another good package
> for ML (maximum likelihood) missing data imputation? I have seen
> several recommendations online, but I wonder which one is currently
> the "reigning champ."
The "Multivariate" and the "SocialSciences" task views have sections about
missing data
http://CRAN.R-project.org/view=Multivariate
http://CRAN.R-project.org/view=SocialSciences
Additionally, I can recall that there is the Amelia II package:
http://CRAN.R-project.org/package=Amelia
and potentially others.
hth,
Z
Uwe Ligges
Oh yes, indeed.
Please ignore my former message, I haven't looked at the license
carefully enough. And since the license is rather special (not GPL'ed as
I assumed in my message), you cannot take over maintainership without
negotiation with the former maintainer, I fear.
Best,
Uwe Ligges
Ted Harding
On a point of information: The licence in question:
License: The software may be distributed free of charge and used
by anyone if credit is given. It has been tested fairly
well, but it comes with no guarantees and the authors
assume no liability for its use or misuse.
is verbatim from Joe Shafer's original licence for his NORM.
He used exactly the same wording for his packages CAT, MIX and PAN.
These were originally written in S, with FORTRAN code for many of
the functions. The various people who have ported these to R have
simply copied these words into the R packages. See (if you have
the packages installed)
library(help=cat)
library(help=norm)
library)help=mix)
I may have some comments about this "removed from CRAN" issue later,
but I need to think about them first ...
Best wishes to all,
Ted.
--------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.H...@manchester.ac.uk>
Fax-to-email: +44 (0)870 094 0861
Date: 17-Jul-09 Time: 13:38:23
------------------------------ XFMail ------------------------------
Ted Harding
On 17-Jul-09 12:38:27, Ted Harding wrote:
> On a point of information: The licence in question:
>
> License: The software may be distributed free of charge and used
> by anyone if credit is given. It has been tested fairly
> well, but it comes with no guarantees and the authors
> assume no liability for its use or misuse.
>
> is verbatim from Joe Shafer's original licence for his NORM.
> He used exactly the same wording for his packages CAT, MIX and PAN.
> These were originally written in S, with FORTRAN code for many of
> the functions. The various people who have ported these to R have
> simply copied these words into the R packages. See (if you have
> the packages installed)
> library(help=cat)
> library(help=norm)
> library)help=mix)
>
> I may have some comments about this "removed from CRAN" issue later,
> but I need to think about them first ...
>
> Best wishes to all,
> Ted.
While Joe Shafer's MI software web page
http://www.stat.psu.edu/~jls/misoftwa.html
as cited in the R help pages, still exists (though apparently still
dating from 1999), only the Windows versions of NORM, CAT, MIX and PAN
are still accessible. The link to the Unix versions:
"S-PLUS for Unix:" http://www.stat.psu.edu/~jls/splunix.html
no longer leads anywhere (though it still did only a few years ago).
I don't know whether it has been moved, and can be found elsewhere,
ot whether Shafer has withdrawn it.
The Windows versions would be compiled code. The Unix versions were
source code and needed to be compiled. If Shafer has withdrawn the
source-code versions, this might have implications for use (these
underlie the S-Plus "missing" library).
Does anyone have information about this?
Ted.
--------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.H...@manchester.ac.uk>
Fax-to-email: +44 (0)870 094 0861
Date: 17-Jul-09 Time: 13:54:45
Gabor Grothendieck
Ted Harding
> On Fri, Jul 17, 2009 at 8:54 AM, Ted
> Harding<Ted.H...@manchester.ac.uk> wrote:
>> Follow-up:
>>
>> On 17-Jul-09 12:38:27, Ted Harding wrote:
>>> On a point of information: The licence in question:
>>>
>>> License: The software may be distributed free of charge and used
>>> _ _ _ _ _well, but it comes with no guarantees and the authors
>>> _ _ _ _ _assume no liability for its use or misuse.
>>>
>>> is verbatim from Joe Shafer's original licence for his NORM.
>>> He used exactly the same wording for his packages CAT, MIX and PAN.
>>> These were originally written in S, with FORTRAN code for many of
>>> the functions. The various people who have ported these to R have
>>> simply copied these words into the R packages. See (if you have
>>> the packages installed)
>>> _ library(help=norm)
>>> _ library)help=mix)
>>>
>>> I may have some comments about this "removed from CRAN" issue later,
>>> but I need to think about them first ...
>>>
>>> Best wishes to all,
>>> Ted.
>>
>> While Joe Shafer's MI software web page
>>
>>
>> as cited in the R help pages, still exists (though apparently still
>> dating from 1999), only the Windows versions of NORM, CAT, MIX and PAN
>> are still accessible. The link to the Unix versions:
>>
>>
>> no longer leads anywhere (though it still did only a few years ago).
>
> Its archived here:
> http://web.archive.org/web/20050212143644/http://www.stat.psu.edu/
> ~jls/splunix.html
Thanks, Gabor!
Ted.
--------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.H...@manchester.ac.uk>
Fax-to-email: +44 (0)870 094 0861
Date: 17-Jul-09 Time: 14:27:26
------------------------------ XFMail ------------------------------
KathyKlein
Hello,
I also used the norm-package for data with missing values. Are there no
possibilities to fix it (its not under GPL?), or is here anyone able to get
the negotiation for the maintainership for this package?
Another question: I have read, taht "mice" would be an alternative instead
of using "norm". Is it also possible to the EM-Algorithm for imputation of
missing values in multivariate normal data? Or are there better
possibilities?
(Besides:the package "mvnmle" is not possible for me, because my data matrix
is too big...)
I would be pleased to get anwers from you!
Katharina Klein
--
View this message in context: http://www.nabble.com/Package-norm-has-been-removed.-What-to-use-for-Maximum-Likelihood-Missing-Data-Imputation--tp24531976p24718866.html
Sent from the R help mailing list archive at Nabble.com.