topic relationship representation in SAS Programming

relationship representation

HeatherNewton — Sat, 29 Jul 2023 01:26:44 GMT

HiI have 40 sas programs and over 100 input and output sas data sets (as source and output). What is a good way to present these relationships? I find it daunting to keep using search function to know what is input and output of each sas programs. Is there any free tools output I can easily represent may be in a diagram or flowchart? what would an expert do here to easily understand the relationship?

Re: relationship representation

ballardw — Sat, 29 Jul 2023 03:58:00 GMT

You likely have access to SAS Enterprise Guide even if not using it.

You would have to create project(s) and build the relationships between the program and sets.

Or if you have SAS/Graph you may have the Network Visualization Workshop available, check your online documentation.

You would build data sets to indicate what you need as node and links

Re: relationship representation

Kurt_Bremser — Sat, 29 Jul 2023 05:47:29 GMT

If these are automated programs, the scheduler might have the dependencies defined, and you could get a graph from there.

Re: relationship representation

Patrick — Sat, 29 Jul 2023 09:20:25 GMT

I've never seen a satisfying solution for SAS code when it comes to column level data lineage. Anything I've seen required a lot (and I mean A LOT) of work for creation and ongoing maintenance.

For only a 100 programs and table level lineage things might be manageable though.

There are data lineage tools but they all require metadata. If you're using SAS DI Studio then such metadata would exist - but else it likely doesn't.

I have also seen attempts to create such mappings by scanning SAS logs - but with SAS allowing for dynamic code using macros this was never 100% correct and always required manual checks and corrections.

Just for documentation purposes what you could be doing is to create a matrix. Something like:

I: Input

O: Output

...but of course like for any documentation it then also requires maintenance whenever you change a program.