However, when I plug in my data and let it run, I occasionally will get back result with some of the DMUs having theta greater than one. This has happened on multiple occasions. That should be impossible in DEA. Is this a known bug with a solution or update?
[USFS]
Gregory E Frey, PhD
Research Forester
USDA Forest Service
Southern Research Station, Forest Economics and Policy
When you specify the output orientated DEA option, the efficiency scores
will be greater or equal to 1. The program is coded like this manner because
it tells us that certain DMUs with higher scores compared to others
should've resulted in more output with the same input level for output
oriented DEA. If you want to get the scores less than 1 and greater or equal
to 0, please simply take inverse of them.
And I updated the *.ado files in the following site.
So, please visit the following site and download the -*.ado-, -ldea.do-
files from Svn/trunk folder.
You can make your ID and come in to the project. Then, we will alert the
revision.
Delete or rename all the previous version of *.ado file.
Save the files in your stata program folder; under the./personal
Open your data set.
Run "do ldea.do" using Stata once(this is required only once) or save
ldea.mlib file in your directory ". /personal" folder.
If stata still does not recognize the * command, you need to make sure
that all the .ado and .mlib files are in right folder.
By default, Stata for Window system uses the c:\ado directory for
user-written ado files. You can list Stata's system directories by executing
the -sysdir- command:
If Stata does not recognize your command, please save all the .ado and .mlib files in the "c:\ado\personal\" folder. If you don't have this
folder, you can make it.
You can use the -sysdir set- command. Suppose you store the user written ado
files in the c:\ado\plus directory. You would then need to run the following
command:
.sysdir set PLUS "c:\ado\plus"
For your case, use
.sysdir set PERSONAL "c:\ado\personal"
Stata will then find user-written commands in the c:\ado\plus or
c:\ado\personal directory.
And check if the folder have all the *.ado files.
If there is any further question, let me know.
Choonjoo.
From: G Frey [mailto:freyge00@users.sf.net]
Sent: Saturday, May 16, 2015 7:44 AM
To: bloom.rampike@gmail.com
Subject: [deas:discussion] Data Envelopment Analysis theta > 1
However, when I plug in my data and let it run, I occasionally will get back
result with some of the DMUs having theta greater than one. This has
happened on multiple occasions. That should be impossible in DEA. Is this a
known bug with a solution or update?
[USFS]
Gregory E Frey, PhD
Research Forester
USDA Forest Service
Southern Research Station, Forest Economics and Policy
Thanks for your help. With the updated files you provided in your link, the dea function appears to be functioning correctly.
FYI, in the original question, I was using the default option, which should be the input-oriented model, and I was getting mostly efficiencies less than one, but about 5-10% of them coming back greater than one.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
When runing "do ldea.do" command in Stata after I downloaded all the .ado files, Stata writes the following message
(note: file c:\ado\personal\ldea.mlib not found)
file c:\ado\personal\ldea.mlib could not be opened
(3 lines skipped)
r(603);
My question is: Where can I find ".mlib" file to download in right in folder on my comp?
And I also have a second question: when runing dea command in results I see that the sum of lambda less than 1. That should be impossible, this condition (sum of lambdas = 1) DEA's key point. Have you ever had such strange? How can I fix this moment?
Thank you in advance!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear Altynay,
I had the same error and fixed it.
I found out that my ado/ folder did not have a personal/ folder in it. I created the personal folder manually and ran ldea.do again and the error was gone.
You can try that.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello, I am using a data envelopment analysis module, dea, authored by Ji and Lee. http://www.stata-journal.com/article...article=st0193
However, when I plug in my data and let it run, I occasionally will get back result with some of the DMUs having theta greater than one. This has happened on multiple occasions. That should be impossible in DEA. Is this a known bug with a solution or update?
[USFS]
Gregory E Frey, PhD
Research Forester
USDA Forest Service
Southern Research Station, Forest Economics and Policy
p: 919-549-4025
GregoryEFrey@fs.fed.usGregoryEFrey@fs.fed.us
P.O. Box 12254
Forestry Sciences Laboratory
3041 E. Cornwallis Rd.
Research Triangle Park, NC 27709
www.fs.fed.ushttp://www.fs.fed.us/
[http://fsweb.wo.fs.fed.us/esignature/images/usda-logo.png]http://usda.gov/[Twitter]https://twitter.com/forestservice[Facebook]http://facebook.com/USDA
Caring for the land and serving people
Dear Gregory E Frey,
When you specify the output orientated DEA option, the efficiency scores
will be greater or equal to 1. The program is coded like this manner because
it tells us that certain DMUs with higher scores compared to others
should've resulted in more output with the same input level for output
oriented DEA. If you want to get the scores less than 1 and greater or equal
to 0, please simply take inverse of them.
And I updated the *.ado files in the following site.
So, please visit the following site and download the -*.ado-, -ldea.do-
files from Svn/trunk folder.
http://sourceforge.net/p/deas/code/HEAD/tree/trunk/
Or search "deas" in https://sourceforge.net https://sourceforge.net site.
You can make your ID and come in to the project. Then, we will alert the
revision.
Delete or rename all the previous version of *.ado file.
Save the files in your stata program folder; under the./personal
Open your data set.
Run "do ldea.do" using Stata once(this is required only once) or save
ldea.mlib file in your directory ". /personal" folder.
If stata still does not recognize the * command, you need to make sure
that all the .ado and .mlib files are in right folder.
By default, Stata for Window system uses the c:\ado directory for
user-written ado files. You can list Stata's system directories by executing
the -sysdir- command:
For example, in my computer,
.sysdir
If Stata does not recognize your command, please save all the .ado and
.mlib files in the "c:\ado\personal\" folder. If you don't have this
folder, you can make it.
You can use the -sysdir set- command. Suppose you store the user written ado
files in the c:\ado\plus directory. You would then need to run the following
command:
.sysdir set PLUS "c:\ado\plus"
For your case, use
.sysdir set PERSONAL "c:\ado\personal"
Stata will then find user-written commands in the c:\ado\plus or
c:\ado\personal directory.
And check if the folder have all the *.ado files.
If there is any further question, let me know.
Choonjoo.
From: G Frey [mailto:freyge00@users.sf.net]
Sent: Saturday, May 16, 2015 7:44 AM
To: bloom.rampike@gmail.com
Subject: [deas:discussion] Data Envelopment Analysis theta > 1
Hello, I am using a data envelopment analysis module, dea, authored by Ji
and Lee. http://www.stata-journal.com/article...article=st0193
However, when I plug in my data and let it run, I occasionally will get back
result with some of the DMUs having theta greater than one. This has
happened on multiple occasions. That should be impossible in DEA. Is this a
known bug with a solution or update?
[USFS]
Gregory E Frey, PhD
Research Forester
USDA Forest Service
Southern Research Station, Forest Economics and Policy
p: 919-549-4025
GregoryEFrey@fs.fed.usGregoryEFrey@fs.fed.us GregoryEFrey@fs.fed.us
P.O. Box 12254
Forestry Sciences Laboratory
3041 E. Cornwallis Rd.
Research Triangle Park, NC 27709
www.fs.fed.ushttp://www.fs.fed.us/ http://www.fs.fed.us
[http://fsweb.wo.fs.fed.us/esignature/images/usda-logo.png]http://usda.gov/
http://usda.gov
[Twitter]https://twitter.com/forestservice[Facebook]http://facebook.com/USDA
Caring for the land and serving people
Data Envelopment Analysis theta
http://sourceforge.net/p/deas/discussion/help/thread/737d0d22/?limit=25#799 c > 1
Sent from sourceforge.net because bloom.rampike@gmail.com
bloom.rampike@gmail.com is subscribed to
http://sourceforge.net/p/deas/discussion/help/
http://sourceforge.net/p/deas/discussion/help
To unsubscribe from further messages, a project admin can change settings at
http://sourceforge.net/p/deas/admin/discussion/forums. Or, if this is a
mailing list, you can unsubscribe from the mailing list.
Thanks for your help. With the updated files you provided in your link, the dea function appears to be functioning correctly.
FYI, in the original question, I was using the default option, which should be the input-oriented model, and I was getting mostly efficiencies less than one, but about 5-10% of them coming back greater than one.
Dear Choonjoo,
When runing "do ldea.do" command in Stata after I downloaded all the .ado files, Stata writes the following message
My question is: Where can I find ".mlib" file to download in right in folder on my comp?
And I also have a second question: when runing dea command in results I see that the sum of lambda less than 1. That should be impossible, this condition (sum of lambdas = 1) DEA's key point. Have you ever had such strange? How can I fix this moment?
Thank you in advance!
Dear Altynay,
I had the same error and fixed it.
I found out that my ado/ folder did not have a personal/ folder in it. I created the personal folder manually and ran ldea.do again and the error was gone.
You can try that.