topic Re: Is there a Github repo or similar for SUGI? in All Things Community

Is there a Github repo or similar for SUGI?

afiqcjohari — Fri, 17 Mar 2017 08:11:08 GMT

I want to learn more about SAS using SUGI which are available onlines.

However, it seems that it's tedious to follow the tutorial as

1) Copy pasting the code used in SUGI doesn't work as intended i.e not formatted, characters wrong, pagination

Making learning from SUGI more laborious than intended.

As someone used to do statistical programming using R and Python, it makes me wonder how SAS actually still thrived in these days and age.

Re: Is there a Github repo or similar for SUGI?

Kurt_Bremser — Fri, 17 Mar 2017 08:14:41 GMT

@afiqcjohari wrote:

As someone used to do statistical programming using R and Python, it makes me wonder how SAS actually still thrived in these days and age.

Maybe because the old SAS hands are really computer-savy and quite intelligent, which explains their rather high salaries?

Re: Is there a Github repo or similar for SUGI?

afiqcjohari — Fri, 17 Mar 2017 09:30:36 GMT

Not interested in the language war as it's pretty useless without context though it's arguable that R and Python or open source ecosystem hasgained exponential tractions among the younger intelligent computer savvy generations. Facebook Uber and AirBnb use these tools for data science.

Though it's definitely a challenge to replace SAS as it's tied to legacy systems which many companies still use.

But this doesn't answer the question.

Career wise, it does feel like SAS will turn out to be like COBOL or Fortran. I think it's only reasonable to gauge whether investing time and effort to master SAS is a good thing or not.

Re: Is there a Github repo or similar for SUGI?

Kurt_Bremser — Fri, 17 Mar 2017 09:38:13 GMT

Comparing R and Python-based statistics with SAS is not even comparing apples to oranges, IMHO.

SAS is a complete data warehousing system, from ETL to statistics to GUI-based Analytics tools for the non-programmers.

Including the necessary server backends like LAZR, and the necessary metadata infrastructure.

Although the SAS language is at the heart of it, it is only a small part of the picture when you look at data warehousing from an IT manager's POV.

How good is R at delivering a result to the archetypical PHB from a combination of Excel files, legacy text files and databases? In a way that allows the PHB to interact with the data, without any in-depth knowledge of what's behind it technically?

Re: Is there a Github repo or similar for SUGI?

Reeza — Fri, 17 Mar 2017 09:55:50 GMT

One tiny point, SUGI outdates Github (and StackOverflow) by several decades. The issues you're running into are from copying and pasting from PDF files. Paste to a text file first, zap gremlins, and then paste it into SAS.

Re: Is there a Github repo or similar for SUGI?

Reeza — Fri, 17 Mar 2017 09:58:56 GMT

The programming language doesn't matter, the problem solving skills are the same. Choosing to learn is ultimately your choice.

Re: Is there a Github repo or similar for SUGI?

afiqcjohari — Fri, 17 Mar 2017 10:01:14 GMT

Since SUGI is outdated, is there a SUGI equivalent which is modern like? Something similars to Jupyter/ipython Notebook maybe?

SUGI should have been more reproducible in my opinion.

Re: Is there a Github repo or similar for SUGI?

Reeza — Fri, 17 Mar 2017 10:19:44 GMT

SUGI is SAS User Group International - an annual conference that occurred starting in the 70s (?)

SAS Global Forum is the equivalent, papers and presentations are indexed on SAS support site and many papers have separate code files as text files.

A bit old fashioned but one of the best resources is Ron Codys book, learn by example. Or see the UCLA online SAS tutorials which are annotated.

https://www.sas.com/store/books/categories/getting-started/learning-sas-by-example-a-programmer-s-guide/prodBK_60864_en.html

http://stats.idre.ucla.edu/sas/

Re: Is there a Github repo or similar for SUGI?

afiqcjohari — Fri, 17 Mar 2017 10:47:30 GMT

I beg to differ, the language does matter. It breaks or makes your career. Choosing to learn is easy if time is abundant but not so as I think time is a luxury.

Here's a good anecdote.
https://www.r-bloggers.com/on-programming-languages-why-my-dad-went-from-programming-to-driving-a-bus/

Re: Is there a Github repo or similar for SUGI?

Ksharp — Fri, 17 Mar 2017 10:48:17 GMT

As time goes by, you will know SAS is better than R or Python .
SAS is not just a simple program language, unlike other language Jave,Dephi,C,Object C.....
SAS 's heart or backup is Statistical and Mathematics Theory . 
Do you think Statistical and Mathematics would obsolete ?

Re: Is there a Github repo or similar for SUGI?

Kurt_Bremser — Fri, 17 Mar 2017 11:37:30 GMT

@afiqcjohari wrote:
I beg to differ, the language does matter. It breaks or makes your career. Choosing to learn is easy if time is abundant but not so as I think time is a luxury.

Here's a good anecdote.
https://www.r-bloggers.com/on-programming-languages-why-my-dad-went-from-programming-to-driving-a-bus/

From the end of that anecdote:

"Someone needs to take a look at this. End of rant."

With SAS, that "someone" sits in Cary, NC. With some open source systems, that someone often turns out to be noone.

@Reezacan probably give you an account of her experiences with R vs. SAS.

While languages like COBOL are really in the state of "smelling funny", the SAS ecosystem is still evolving. Just take a look at the many additions from 9.2 to 9.4 (proc ds2, among others).

Re: Is there a Github repo or similar for SUGI?

Kurt_Bremser — Fri, 17 Mar 2017 11:40:24 GMT

@afiqcjohari wrote:
Choosing to learn is easy if time is abundant but not so as I think time is a luxury.

When you're through learning, you're through. It's not a matter of time at hand, it's a state of mind. I always learned new languages (with the exception of the first one, Pascal) under time pressure, as I had to start fixing something NOW.

Re: Is there a Github repo or similar for SUGI?

ChrisHemedinger — Fri, 17 Mar 2017 14:30:01 GMT

In reply to the original question: is there a GitHub repo for SAS Global Forum? There isn't a repo specifically for the conference, but many individual authors do share projects on GitHub.

SAS has a GitHub account at https://github.com/sassoftware. SAS employees and some other contributors have created projects that they share on GitHub, many of which will be discussed at SAS Global Forum.
Other users/presenters may use GitHub to share supporting code for their papers. I've done this many times. Here are some examples from @Jiangtang. You can search LexJansen.com for "GitHub" and find many more (thank you @Lex_SAS!).
SAS R&D developers share many examples here on the community, and they use GitHub as a source code repository. Here's an example from SAS Enterprise Miner. Here's a video of @WendyCzika (and yours truly) discussing this effort.

Chris

Re: Is there a Github repo or similar for SUGI?

Quentin — Fri, 17 Mar 2017 14:21:26 GMT

I like this idea of a repository for SUGI / SGF / RUG code. @Lex_SAS made it a much better world when he provided a way for the user community to search the universe of UG papers.

But in terms of sharing code associated with those papers, the method varies dramatically among authors ("see my website", "see my github page", "see my sascommunity page", "see code in the appendix," "email me for code", "my company won't let me share the code but thanks for letting me present", etc.)

Maybe it's worth considering options for storing code relating to papers, in a useful way that is searchable etc. Allow authors to submit their code along with their paper (or perhaps let the author upload the code themselves).

This would also be useful for code associated with SAS publishing books. I think currenlty, accessing that code usually means downloading a .zip file from some site.

Re: Is there a Github repo or similar for SUGI?

ChrisHemedinger — Fri, 17 Mar 2017 14:48:53 GMT

Thanks @Quentin. We have used GitHub as a place for some book examples:

My book: Developing Custom Tasks for SAS Enterprise Guide makes up most of the content on my account.
Kevin Smith's new book about SAS Viya and Python has lots of Jupyter notebook examples.
SAS programming for R users is a free SAS course, with all course notes, data, and code on GitHub

That said, we have to recognize that GitHub is not a site that any of us owns, and it's subject to its own terms and conditions that we don't control. For that reason, we don't usually make such materials available exclusively on GitHub -- we also put them other places (like those ZIP files you mention).

Re: Is there a Github repo or similar for SUGI?

Quentin — Fri, 17 Mar 2017 17:04:28 GMT

Thanks @ChrisHemedinger. I recognize that you are one of the leaders in posting SAS examples and tools to GitHub (both from your book, and your blog). So thanks for that.

I think it's a reasonable question for SGF and Regional User Group committees to consider:

Right now, most UG conferences do a good job of archiving papers, and making them accessible and searchable. Would there be a benefit to having a standard way of archiving the related code, and make it accessible and google-able?

Well, I just wandered over to

http://www.lexjansen.com/cgi-bin/xsl_transform.php?x=sgf2016&c=sugi

And see that there are a good hanful of papers from last year that have the option to download a data file. And the papers and data files are of course also available at the source:

http://support.sas.com/resources/papers/proceedings16/

So that's great. Still, maybe if this code was stored in a more code-friendly format that was searchable, more people would find it.

That said, I think it's great that SGF has this option for authors to upload code. Browsing through the RUGs, it looks like there are very few papers where authors have attached a data file. I think anything UGs can do to encourage authors to submit code along with their paper, and host that code and related files along with the papers, would be a benefit to the SAS community.

Re: Is there a Github repo or similar for SUGI?

Lex_SAS — Fri, 17 Mar 2017 18:27:46 GMT

Yes, at lexjansen.com I try to make additional resources available with the papers. But it is a manual process for adding this metadata.

For years, I had a banner inviting people to contact me if they wanted any code or presentation added to their paper link.

The response to that was minimal.

I'm always open to other suggestions.

Having people upload their own code would be ideal, but adds a lot of management (user management, security, ...)/

Lex

Re: Is there a Github repo or similar for SUGI?

Reeza — Fri, 17 Mar 2017 19:15:24 GMT

I think this is a personal work style issue rather than just post SUGI code on GitHub.

The conferences/paper are all volunteer written and there is no organization that I know of that would take on converting all that code to GitHub.

One of the best advices I received when starting to learn a language was to create my own code snippets and keep those in a handy reference document. There are others out there, but building your own is a useful learning exericise.

I keep my snippets on GitHub and there are other public SAS repositories as shown by @ChrisHemedinger

https://gist.github.com/statgeek

Re: Is there a Github repo or similar for SUGI?

Reeza — Fri, 17 Mar 2017 19:22:29 GMT

Last comment, R/Python are open source, SAS is not.

I like that I get paid for my work, and paid well, so making 'everything' open probably isn't the road I'm going to head down. Intellectual property is a great idea, colloboration is a great idea, open source is a great idea, but I also need to pay my pharmacy so I have enough drugs to function semi-normally. Drugs are expensive.

FYI - This is starting to be a larger issue in open source communities.

Re: Is there a Github repo or similar for SUGI?

Quentin — Sat, 18 Mar 2017 02:26:37 GMT

Agree @Reeza, wan't suggesting that someone go through past proceedings and extract all the code. And to be honest, it's been far too long since I've submitted a paper for a UG, so I can't remember how it worked.

But I can imagine a process where, when authors submit their final paper, they were also encouraged to submit a separate file with code (and perhaps data). So the UG volunteers responsible for managing publication of the proceedings would have the papers and the associated code file, allowing them to post both. The papers could be posted as they are now, but the code could be in some sort of code-friendly repository. Not necessarily designed for collaboration like github, but somehwhere where code would be google-able, copy-and-paste-able, and perhaps even discussable.

I think if the authors uploaded the separate code file along with their paper, it wouldn't be a lot more work to post the code files somewhere useful. And if it were done in a systematic way, hopefully it would mean less manual work for @Lex_SAS. : )