topic Re: Renaming a Dataset in SAS Programming

Renaming a Dataset

Kody_devl — Fri, 12 Aug 2016 14:26:16 GMT

Hi,

I am using SQL to build my datasets and get "warnings" if the name of the modified new dataset is the same as the original dataset.

For example:

Dataset Old =

ColA, ColB

Dataset New =

ColA, ColB, ColC

(one attribute added)

When I call the new dataset the same name as the name of the old dataset, I get a "warning" in the error log.

What is a good and simple technique to build my dataset out and so that

1. I end up with only one dataset and

2. The result has the name of the Original dataset (even if I have modified the dataset 5 or 6 times) and

2. I can eliminate all of the warnings?

A simple example would help.

(when I research, I find much on renaming attributes but not much on datasets)

Renaming a Dataset

Kody_devl — Fri, 12 Aug 2016 14:23:15 GMT

I am using SQL to build my datasets and get "warnings" if the name of the modified new dataset is the same as the original dataset.

For example:

Dataset Old =

ColA, ColB

Dataset New =

ColA, ColB, ColC

(one attribute added)

When I call the new dataset the same name as the name of the old dataset, I get a "warning" in the error log.

What is a good and simple technique to build my dataset out and so that

1. I end up with only one dataset and

2. The result has the name of the Original dataset (even if I have modified the dataset 5 or 6 times) and

2. I can eliminate all of the warnings?

A simple example would help.

(when I research, I find much on renaming attributes but not much on datasets)

Re: Renaming a Dataset

Reeza — Fri, 12 Aug 2016 14:28:34 GMT

If your using multiple steps, use names that are unique to each step and then make a final dataset. You can use proc datasets to both remove extra data sets and to rename the final dataset if you'd like.

Re: Renaming a Dataset

RW9 — Fri, 12 Aug 2016 14:32:11 GMT

The warning you get is telling you that writing to the same dataset that you read from will mean you cannot roll the changes back. Its an SQL implementation thing. You can put:

proc sql undopolicy=none;

...

That will remove the wraning. However you may be better off just using datastep language - without seeing your process I can't really advise, but you can merge various datasets and do logic in one step:

data want;
  set old:;
  colc="Hello World";
run;

This would bring in all datasets in work with the prefix old, and create a new variable in the output want dataset.

Re: Renaming a Dataset

Haikuo — Fri, 12 Aug 2016 14:35:33 GMT

However NOT recommended, You can opt to data step to get rid of the warning.

data old;
set old;
run;

But if anything goes wrong (e.g. power outage), then you may lost your original data.

@Reeza's suggestion is dead on. Name it differently for every change you make, then using Proc dataset to clean the house, doing that guarantee you easy trouble-shoot and better approach of programming.

Re: Renaming a Dataset

Kody_devl — Fri, 12 Aug 2016 17:29:31 GMT

This Works couple with dropping the Old Dataset.

Thank you

Re: Renaming a Dataset

MMMIIIIKKKKEEEE — Fri, 12 Aug 2016 18:14:56 GMT

Cody_Devl,

I see that you got your answer and I'll bet that you have already implemented it. Great!

I have another idea that will address requirements #2 and #3 that may come in handy some day. Consider creating Generation Data Groups for the data sets you want to modify. For example:

proc datasetslibrary=sgflib nolist;

modify ColA(genmax=10);

modify ColB(genmax=10);

run;

quit;

Then, when you modify ColA or ColB, SAS will save the old copies as ColA#001, ColB#001, respectively and keep the modified versions.

Don't take my word for it; check it out!