Analytics and insights software provider SAS has released a new SAS Viya Workbench to give developers a modern, lightweight, scalable environment to work in - supporting SAS and Python languages with R support to come by the end of 2024. The Workbench also allows developers their choice of Visual Studio Code and Jupyter notebooks.
SAS has provided the world with number-crunching software since the 1970s and while a major strength of SAS is its strong backward compatibility, it's certainly been no slouch when it comes to embracing moderm major innovations - that includes the cloud, as well as open source technologies and languages.
The SAS language has been built into the product from the start, evolving over decades. It's a powerful language that is understated in its subtlety. For instance, the SAS DATA STEP allows you to move data from one location to another in a few lines of code. You can perform transformations and formatting with a few more lines of code. If you have a load of data in, say, an Oracle database, and you want it in a Teradata data warehouse, you only need three lines of code. And with a couple of extra lines, you can take advantage of the vendor's parallel capabilities. There are a lot of companies performing data movement at a lot more cost and with a lot more complexity. Similarly, you can pull data from a database table on a remote server, recode and clean it, transform it into tabular data, calculate rates, and output a multi-tab Excel spreadsheet for public consumption with not a lot of effort.
SAS is in use in many highly-regulated industries like medicine and pharmaceutical (as well as other industries of all types), and the company's language and features are thoroughly documented and tested, including all the right regulatory qualifications and validation. That's a huge time and effort the company has put in which guarantees every SAS procedure is statistically sound and error-free. That's an assurance you simply do not get from the bulk of other analysis tools available in the world today.
{loadposition david08}
From that point of view, using the SAS language and SAS platform is a solid, safe, choice that brings efficiency and powerful functionality, as well as a guaranteed long-term lifetime.
In fact, during the company's major annual SAS Innovate conference last month, co-founder and CEO Dr. Jim Goodnight took to the stage to demonstrate the new SAS Viya Workbench and used code of his own, from the 1980s, to demonstrate.
OMG Dr Jim Goodnight himself is going to demo coding in the new SAS Viya Workbench #SASInnovate pic.twitter.com/5UqrZ5YAnx
— davidmwilliams (@davidmwilliams) April 17, 2024
And here’s the classic cowboy hat - code by Dr Jim Goodnight written in the 1980s and running in 2024 on the modern SAS Viya Workbench #SASInnovate pic.twitter.com/vCFeuRO17o
— davidmwilliams (@davidmwilliams) April 17, 2024
This takes us to SAS Viya Workbench itself - while the company has its traditional SAS enterprise-grade on-premises or hosted SAS 9.4 software, it also comes in a cloud flavour with SAS Viya, or Visual Analytics. Viya is a SaaS-based offering, but also brings a major focus on the AI lifecycle, including modelling and algorithms. It has greater always-on capability, supports more users and transactions, and brings a slew of other more powerful features. And, as per Dr. Goodnight, it's still backwards-compatible, code-wise.
Of course, you might not want to work in the SAS language, and the company has worked hard to bring its analytic capability and strength to organisations, meeting them where they are.
SAS Viya Workbench
The most recent fruits of this is SAS Viya Workbench, a flexible, lightweight developer environment for building AI models no matter if you prefer Python to SAS - or, coming later this year, R - and no matter if you prefer Visual Studio Code or Jupyter notebooks.
Whatever your preference, SAS Viya Workbench is now generally available and has been targeted to developers and modellers as a self-service, on-demand computer environment for data preparation, exploratory data analysis, and developing analytical and machine learning models.
In fact, even if you have existing Python code, you'll find performance gains by bringing it to SAS and its own custom Python libraries. In one example on-stage at SAS Innovate, the team showed a complex algorithm that ran in 45 minutes using the open-source Python sklearn.svm.SVC maths library. By making one change only - adjusting the import to sasviya.ml.svm.SVC - the code completed in nine seconds.
"We've got 47 years of experience doing this stuff," said SAS principal software developer Joseph Henry. Dr. Goodnight joked, "open source may be free but compute is not."
SAS Viya Workbench is a flexible, scalable and efficient development environment that is on-demand, self-provisioning and self-terminating with minimal IT support. The dedicated analytical environment features customisable CPU/GPU compute power to match the needs of the project. Models and other results can be leveraged in SAS Viya for data management, governance and operational deployment.
SAS Viya Workbench will initially be available through the Amazon AWS Marketplace in Q2, with future plans for additional supported cloud providers and a software-as-a-service deployment option.
That's not all; "there has never been a better time in history to be a developer," said SAS global AI strategy lead Marinela Profi.
Profi went on to demonstrate additional enhancements the company has made to the developer experience including SAS Viya Copilot, customised generative AI models, and SAS Data Maker, to create synthetic data as needed.
SAS Viya Copilot
SAS Viya Copilot naturally slips into a SAS developer's workflow and provides helpful natural-language prompted guidance. For example, if you have a set of data you can ask SAS Viya Copilot to create a trend analysis or time series forecast. You could ask it what would be a good way to analyse and plot data, and to bundle it all up into a dashboard to share with executives. You can ask SAS Viya Copilot to brainstorm with you and come up with ideas on how you can do more with your data.
Going further, SAS Viya Copilot also allows you to make your own personal series of copilots for different purposes. You set it up by answering some simple questions, and you gain an LLM-agnostic copilot that will answer questions on the data and industry you provide it with. The copilot runs on your own SAS Viya server and is not shared with any other organisation. It preserves your company's data governance.
"This is the future of data science development, AI-powered developers. I'm excited to be on this journey with you," Profi said.
SAS Data Maker
And, next, SAS Data Maker is available currently in private preview, with general availability to come. This is "your gateway to seamless synthetic data generation," Profi said. SAS Data Maker helps users generate synthetic data in a transparent and trustworthy way. It comes with out-of-the-box tools to allow users to validate and compare the distribution of variables among the real and synthetic data while safeguarding sensitive information. It can eliminate the need for costly third-party data collectors, to help provide you with realistic data without the usual cost or effort required. Its output is not proprietary; SAS Data Maker can generate simple CSV files for use in whatever application, system, or tool you like.
Is programming a creative activity or a scientific one? I’ve always felt while there are engineering principles, fundamentally it’s creative. At #SASInnovate @MarinelaProfi gave a really awesome passionate talk on how SAS innovations for devs lets us focus on creativity again. pic.twitter.com/ADbfkUvND7
— davidmwilliams (@davidmwilliams) April 17, 2024
It doesn't end there ... Interested in learning SAS?
Here are some excellent free SAS tutorials and courses to get you started. You will need a free sas.com account.
- Programming 1 essentials (21 hours)
- Statistics 1 introduction to ANOVA, regression, and logistic regression (21 hours)
You can also try for a free SAS account using SAS OnDemand free for academics and SAS Viya for learners. These programs may have differing eligibility requirements.
Also check out the SAS academic portal if you have an .edu address.
Here's my own rendition of Dr. Goodnight's classic cowboy hat using SAS OnDemand:
