Greetings! Again.
Hi everyone. I'm sorry I haven't been around in a while but life happened and I was hoping that the SASInstitute subreddit would be revived. It doesn't seem to have been. But people are still talking about the company. So here's an interesting snippet from 4 months ago
SAS has been around for a long time. It was developed in the 70s on the mainframe. The Language with PROC and DATA step look like it.
Many of us older folks did SAS back in the day. I did SAS in school on the mainframe in the 1980s and on unix systems at work in the 1990s - 2010s, in addition to doing a bunch of other different stuff.
Doesn't mean we like it when we've used it for so long. But here's what some people don't know. SAS is used more for data movement and transformations, along with putting together 'reports' than it is used for analytics. Back in the mainframe days, a lot of business apps were written in COBOL. I'm one of those culprits. But for non-programmers, SAS was a 4GL that was much easier to code than COBOL.
SAS DATA STEP let you move files from one spot to another in a few lines of code. It let you do transformations and formatting with a few more lines of code. So there is a lot of SAS from the 1980s -2000s that are not doing analytics at all. They are doing reporting. SAS is still around in these installations because no one wants to rewrite that code.
SAS/Access software on *nix systems is pretty nifty. If you want to get a pile of data on that Oracle source system to your data warehouse system that might be in Teradata or some other parallel database, SAS/Access with DATA STEP in 3 lines of code will get it moved from one system to another. With 5-6 lines of code you can optimize it so it takes advantage of the database vendors parallel capabilities. This gets used a ton in places I've been. Again, no analytic work here. Just data movement. To do some of this in Python or R is more code, or not very efficient.
All the above is not to praise SAS. I prefer python libraries to do analytic work. Have for many years. But there are a lot of installations with decades of processes that ensure SAS is not going away anytime soon.
SAS has tried to modernize and has made a little bit of progress in that area, but not much in my opinion. For most folks Python / R will be a better fit than SAS nowadays. I left my previous company after doing SAS off and on for over 15 years, and have not touched it at all where I am now.
One thing good about SAS is the support. SAS is expensive and those that say it's expensive probably have not used SAS support very much. I used it all the time. I've had a coworker in the past have a Ph.D. algorithm expert from SAS on the line and emails with my coworker working through a problem with her. SAS support, in my experience has been top notch. I spent a lot of time on the SAS campus with my previous company. We were partnering with them on some things. I'd go say hello to the support folks on campus to put a face to a name.
One thing about SAS, is they give their folks fantastic work conditions on the SAS campus. Pretty much everyone, even the support folks, have their own private office with a door. And these offices are well equipped. I'm thinking that's where some of that money that you are paying to SAS goes. Work life balance on the SAS campus seemed pretty good compared to the company I worked for at the time. 40 hours per week and done.
All that being said, I don't use SAS anymore. I'm using Snowflake, DBT, Fivetran, Python stat/ML libraries and having a good time with that.
Sorry for long comment, but wanted to give my $0.02 on my experience with SAS.
Add comment