topic Re: jackboot macro question in SAS Programming

jackboot macro question

RobertWF — Fri, 10 Feb 2012 21:16:47 GMT

I'm having some difficulty outputing bootstrap sample & confidence interval datasets with the jackboot macro (available for download at http://support.sas.com/kb/24/982.html).

We have Enterprise Guide 4.2 on our PCs in my office, and when I run jackboot (specifically the %boot() and %bootci() macros within jackboot) on my PC everything works fine.

However, when I run jackboot in batch mode via Linux (which we prefer to do vs. running programs on our PCs), only the bootdist output dataset is generated, and the bootdata, bootci, bootpctl, boottrans, and bootstat datasets disappear - I can't find them in my WORK folder, and I'm unable to create permanent datasets (e.g., data boot.bootci; set bootci; run;).

Any ideas what's going wrong? I checked with SAS Support but they couldn't pinpoint what's going wrong in my code.

Thanks,

Robert

Travelers

Re: jackboot macro question

art297 — Fri, 10 Feb 2012 21:33:07 GMT

Can you attach a copy of your log?

Re: jackboot macro question

RobertWF — Fri, 10 Feb 2012 21:59:51 GMT

Absolutely - after Line 8737 towards the end of the log output, you can see that the bootdist, boottran, bootpctl, bootci, bootdist files exist & contain observations, but starting at Line 8742 these same datasets contain zero observations.

Any help is much appreciated!

Re: jackboot macro question

art297 — Fri, 10 Feb 2012 22:12:30 GMT

Robert,

Before spending much time on this, I have a couple of questions.

1. Early in your log is the statment:

You are running SAS 9. Some SAS 8 files will be automatically converted

by the V9 engine; others are incompatible. Please see

http://support.sas.com/rnd/migration/planning/platform/64bit.html

I've never seen that before. Might it be the problem?

2. The line you mentioned 8737 is trying to set bootdata, but that is the first time that bootdata appears in the log

Re: jackboot macro question

Patrick — Sat, 11 Feb 2012 02:29:28 GMT

There are ERROR's in your log like below. You will need to resolve these issues first from top to bottom. The "lost" data sets can be an effect caused by these errors.

There are also notes and warnings which might need investigation (like the note below).

NOTE: Invalid argument to function LGAMMA at line 1357 column 49.

In regards of "it works in EG":

1. EG connects to a SAS Server. Are the servers used for batch and with EG the same? If not: Are the macros used for batch the same (same version)?

2. When running the stuff with EG does it run without Errors the first time after connecting to the server? - The reason for this question: It could be that even though there are errors in the first run you're still creating tables in work which then in a consecutive run might be used and cause the macros to behave differently.

Re: jackboot macro question

RobertWF — Mon, 13 Feb 2012 17:57:50 GMT

Thank you for the suggestions, gentlemen, I did finally find a solution.

Patrick wrote:
There are ERROR's in your log like below. You will need to resolve these issues first from top to bottom. The "lost" data sets can be an effect caused by these errors.
There are also notes and warnings which might need investigation (like the note below).
NOTE: Invalid argument to function LGAMMA at line 1357 column 49. 
In regards of "it works in EG":
1. EG connects to a SAS Server. Are the servers used for batch and with EG the same? If not: Are the macros used for batch the same (same version)?
2. When running the stuff with EG does it run without Errors the first time after connecting to the server? - The reason for this question: It could be that even though there are errors in the first run you're still creating tables in work which then in a consecutive run might be used and cause the macros to behave differently.

The invalid argument errors in the LGAMMA function are a problem, but haven't held up program execution before.

We do use the same server for both EG and batch modes, and it does execute without errors while running the first time in EG.

Robert,

Before spending much time on this, I have a couple of questions.

1. Early in your log is the statment:

You are running SAS 9. Some SAS 8 files will be automatically converted

by the V9 engine; others are incompatible. Please see

http://support.sas.com/rnd/migration/planning/platform/64bit.html

I've never seen that before. Might it be the problem?

2. The line you mentioned 8737 is trying to set bootdata, but that is the first time that bootdata appears in the log

The dataset migration message appears at the top of all of our log files, but don't think that should be an issue.

The bootdata dataset turns out to have been the problem. The jackboot macro does create a dataset named bootdata - however it seems that in batch mode the bootdata dataset is not preserved in the work folder long enough to be saved as a permanent dataset. Once I remove the code:

data boot.bootdata;

set bootdata;

run;

then the other jackboot output datasets (bootci, bootpctl, boottrans, and bootstat) are created & can be output into permanent datasets in my boot library.

So problem solved, thanks for your assistance!

Robert