SAS Intelligent Decisioning, Fairness Monitoring | February and March Release

SAS Intelligent Decisioning, Fairness Monitoring | February and March Release

Show Video

Hello, everyone, and welcome to the SAS Viya Release Highlight Show. I'm Thiago De Souza, and today's show will focus on SAS Intelligent Decisioning. Our expert, Diana Maris, will show us how to reach decisions faster without compromising on governance. We'll kick off the show with our rundown segment. You'll hear about fairness monitoring in SAS Model Manager from Sophia Rowland, custom windows in SAS Analytics for IoT and Event Stream Processing from Aaron White.

Then David Vyck will show us some of his favorite new features, including data documentation in SAS Information Catalog, new sharing options in SAS Visual Analytics, and SAS Studio steps, including quantile regression and tests of proportions. We've got a lot on our plate, so let's dig in. I give you the rundown. Hi, everyone.

Let's talk about what's new in SAS Model Manager. We've just added fairness and bias monitoring into our performance monitoring reports. Modeling teams have been able to assess their models on fairness and bias measures at the time of training, but with this new addition, they can track fairness and bias metrics over time. With this additional information, users can now include measures of model bias when making decisions about when to refresh, replace, or retrain their production models.

The measures we are now tracking includes demographic parity, equal accuracy, equalized odds, equal opportunity, predictive parity, and group unfairness index. To add fairness and bias metrics to performance monitoring reports, teams simply mark which variables they would like to assess for bias. With an understanding of the model's latest performance on bias metrics, teams can be better informed about when to take the appropriate actions.

Hey, everyone. Let's talk about what's new in SAS Event Stream Processing. We recently added custom windows. These are user-created windows that can be reused and shared across multiple projects.

From Studio, you'll see the new custom windows page with a list of all the available custom windows to choose from. We also have support through GitHub link. And creating new and versioning existing windows, you can also import and export custom windows and configuration files. So let me show you how easy it is. Let's pop into an existing project. From the left Windows pane, you'll see Custom.

And below that are all the available custom windows that have been developed so far to pick and choose from. So let's pull in the multiplication custom window and connect source window to it. It's then going to ask us to specify the schema. So we will go to the output schema on the right pane and click the double crosses to automatically import all schema fields from the source window. Go back to the Properties page of the multiplication window. And it is asking us to map the input and output fields.

So simply scroll down to Mappings, select corresponding field you want to map to. Do the same for the output. Save. And now let's test it. Users and teams will use this to dramatically reduce development time, eliminate that pesky, repetitive work that we all hate, and free up time for your team to focus on those other more important priorities. So now we have the data coming in from the source window of our two numbers.

And in the multiplication window, we see our results. And it's just that easy. Thank you, Sophia and Aaron, for today's rundown. Moving right along, it's time to dive into the SAS Viyaverse. Hello, and welcome back to this month's edition, where we will deep dive the new, awesome computing capabilities that have just been added to SAS Viya, and of course, my ongoing quest to convince you to govern your data properly. We are starting things off inside of SAS Information Catalog.

Last time, we took a look at the additional nodes capability, where we can now also add links to our nodes. And now we have also received the ability to add tags on a column level, not just on a data set level, which will enhance the filtering capabilities when you are searching, but also will enable you to easily link things together through the tags on the column level, which is an awesome capability. With the tagging done, let me switch over to SAS Visual Analytics, where there is an enhanced menu on how you can share things with others. Now, let's open up the More menu. And now, instead of just seeing a copy link entry, you also see an additional copy embeddable markup entry.

Let's first go with the copy link one. As you can see, it has been simplified and reduced, but you still get all of the awesome capabilities. So you can check and ensure that the report looks and behaves as you want it to.

In addition, there is also now this copy embeddable markup dialog that enables you to get the direct integration for the SAS Visual Analytics SDK, for example, or get the iframe code, or also additionally add guest access or disable it accordingly. This is a super quick and helpful way to easily share reports with others or help others integrate your report into their applications. Now, the big highlight of this month. Let's switch to SAS Studio and quickly get the two new steps that have been added out of the way, the first one being quantile regression, and the second one being tests of proportioned statistical power, both super helpful steps if you are doing any sort of statistical evaluations. But what I'm really excited about are the new computing capabilities that have been added.

Now, let me show you on the left-hand side. I'm not connected to any CAS libraries at all. I don't even have a CAS session open at the moment. But what I can now do is run the CAS enabled prox also in Compute now, which is an amazing unlock in capabilities. And here, I've just copied an example for an isolation forest from the SAS documentation, and we will be creating our data set in Compute in our work library here.

And then we are going to train this isolation forest model right within Compute. No need to move your data into CAS for this to run, which enables you to make use of the data where it's already residing, and then only push the results up to CAS, for example. And of course, this ran as expected very fast, but it's also a very small data set.

Now, in addition to all the CAS enabled prox becoming available to you inside of Compute, we now also get the ability to call on those same algorithmic capabilities from Python. Let me give you a very small example, but there is a whole much more out there. And if you want me to dive deeper on these capabilities, let me know by leaving a comment down below. Here, I imported from the sasvaya.ml package the forest classifier. And using the common Python help syntax, I get a direct documentation in my log window of all the different things I can do with this specific new object that has been added, meaning I can interact with this forest classifier in a fully Pythonic way.

But in the background, the SAS algorithms are running for me and returning the results to me within Python natively. This really levels up your game as a Python developer to make use of all the awesome algorithmic capabilities that SAS has to offer. There is so much fun stuff to play with. Please, I urge you to check out the documentation of all the awesome new capabilities that this unlocks for you right now. As I said, I'm super excited of all the new algorithmic capabilities that have been unlocked with this release for us. And I cannot wait to see what you will build with it.

See you next time. Bye-bye. That was David Veik with his Into the SAS Viyaverse segment. What do you think about the features he highlighted? Let me know your thoughts in the comments. Intelligent Decisioning automates and optimizes complex decisions using AI, machine learning, and business rules. It enables real-time, data-driven decisions at scale for better efficiency and accuracy.

Our Intelligent Decisioning expert is Dion Ameris for this month's Release Talk interview. Hi, Diona. Welcome back to the show. It's been a while. Hi, Tiago. I'm so excited to be back on the show.

There's so much to catch up on. And it's so good to be here with our audience and you. I'm doing good. I heard you just got back from the Gartner Data and Analytics Summit. Can you tell me how that went and what's new in the SAS Intelligent Decisioning world? Absolutely. Key takeaway for me from the summit was that Gartner views decision intelligence as the next market differentiator for many organizations.

They think that by 2026, up to 70% of organizations will be adopting decision intelligence. Now, that's a big number. And that is because decision intelligence offers the tools to get your return on AI investment.

Organizations are still struggling to get their AI projects into production. And that can be because they're having trouble streamlining the integration of machine learning models into their existing business processes. And there's still a gap of specialized talent as well as that collaboration between the business user who knows the business context and the guardrails that need to be applied. Their collaboration with a data scientist and the IT users is still a little bit fraught with friction. So decision intelligence, by offering a framework to integrate business rules with AI machine learning models and bring them into production within mostly low-code interfaces and cloud environments, makes it much easier for organizations to gain that value from their AI investments. And SAS has been around with a decision intelligence platform for about a decade now.

So it's a very exciting time to be part of this transformative journey that many organizations are going through now. Very cool. You said that 75% of organizations will have adopted intelligent decisioning by 2026. That is a really impressive figure. What I'm curious about is the actual capabilities of intelligent decisioning that these organizations will be taking advantage of. Could you tell me more about that? SAS Intelligent Decisioning focuses on offering a low-code builder for business analysts to be able to integrate AI and machine learning models with business rules and get those into production with the click of a button.

We focus on delivering these decision flows in containerized units in order to be able to take advantage of extensibility, scalability, and other cloud-native features that are available to them. And we do deliver this on all the standard cloud providers out there in the market. And this is a big deal because, as I said, the biggest challenge is to get those into production, get those decisions into production without missing out on scalability and governance. And we actually recently just released a feature for doing that in a scalable manner with a route to live workflow. That's awesome. It sounds like SAS Intelligent Decisioning is all about making you more productive.

The less code, the better. Even I can click a button and get right into it. At the end of your answer there, you mentioned a new workflow.

Can you tell me more about what that workflow entails and how it streamlines things? Yeah, absolutely. So far, we've offered a workflow for governing the development of decisions. But we've noticed a gap in that business users, once they develop these decision flows, they have to go pass that information to IT admins or other stakeholders in order to actually get the asset from a development stage into production.

So now with the workflow, we are actually matching the tasks and steps that organizations need in order to promote assets through environments into production. And we're translating them in easily interpretable tasks within the UI, where users can just point and click and select actions and tasks to complete in order to get that asset directly into production. And we allow organizations to customize those steps in order to match their existing business processes and their existing governance requirements.

For example, being able to set up user permissions and groups for which stakeholders can take which steps. And this, I think, is very powerful because we eliminate that friction in the collaboration between your business and IT. And we get assets into production much faster and doing so in a governed manner. Awesome. Thank you.

That was a very detailed explanation. I'm more of a visual learner. I don't know about you. And I believe you have a demo prepared for us today. So let's take a break from talking about it. And why don't we actually see what it's all about? Here I am in the SAS Viya landing page where I have access to all my AI project assets, from data to machine learning models to decision flows.

As a business analyst, today I need to edit a decision flow to adapt its business rules to the changing needs of my organization. I access the decision flow in Intelligent Decisioning. It is a loan application decision flow. And I've always wanted to have the option to not just develop my decision flows here, but also be able, with the click of a button, to move these assets into production from my design environment. Now, with the workflow that is made available to me directly in Intelligent Decisioning, I can do so. To make my edits, I create a new version of the decision flow.

And because I am an editor in that decision flow, I can assign it a workflow. And when I start that workflow, I have a few parameters that are made available to me by the workflow admin. Now, this decision flow is in development, and I can apply my changes to the business rules.

I am going to increase the requirement for the age of the account. Once I make that change, I can validate and complete my development task. At this point, I can send it for review by a different persona who would be the checker of my decision.

They have access to their tasks also in the home page of SAS Intelligent Decisioning, where they can claim those tasks directly in a task management interface. They can also access the decision directly from there and run some further tests. In this case, scoring the decision with some test data to see how it would behave in the real world.

We can track the paths that the data takes or the transactions are taking through the decision flow with a path tracking and make sure that the results are as expected. Once we have validated that, we can complete the review task. As part of the workflow, I have a requirement to provide an audit document. I can check the box that that step was completed and complete that entire review process. The next step is to make the decision available in a container. This will enable then back-end services to pick up that container and move it across the different development, user acceptance testing, and production environments.

I fill out a few attributes that make it easier to track what am I actually publishing. Now let's have a look at how the workflow steps are evolving. In the SAS Workflow Manager, I can track the instance of my workflow that I had initiated and see which steps are being completed at the moment. When I look at how the workflow was defined, I can see that it's very easy for a business user to just drag and drop components to meet the requirements of business processes in their organization. I can decide which users and groups have access to which tasks, as well as add attributes like that audit document export mandatory check that I just saw my reviewer perform. Some of the components of the workflows have subflows, and this is where I can use the SAS Workflow Manager.

Some of the components of the workflows have subflows. These are very handy if we wanted to reuse steps within a workflow. On the IT side, I have access to visualizing the status of those assets as they are being moved across environments. This is done through seamless integration as we are looking at Argo CD.

This is where I have a comprehensive view of all the environments through which my decision flows are moved, and I can define the automated tasks to do so. Switching back to my SAS Viya stack of applications, I can ensure that the assets are ready for moving into production by performing additional validation. Here I use a template pipeline in SAS Studio where I can just plug and play which versions of the decision I want to compare with what data, and the steps for comparison are already defined for me. The results of that comparison can be analyzed further in a visual analytics dashboard. Now that I did all my validation and progressed through the steps of the workflow, I am ready to make the asset available in production. You saw how, with the help of intuitive, low-code, customizable workflows available in SAS Intelligent Decisioning, business users and IT can collaborate in a frictionless way to get decisioning assets in production much faster while controlling permissions and tracking each step, maintaining governance, and becoming more agile.

Great demo. Seeing what you were talking about previously really helped me understand a little better there. So what's up next for SAS Intelligent Decisioning? What can we see on the horizon? So we are continuing to gather feedback from our customers on the topic of route to live and governance workflows to get decisions into production. We actually have a free webinar available right now for everyone to register and watch on demand on this topic if they want to know more about how to build these workflows and how to apply them in their own use cases. And we'll be at SAS Innovate.

We'll have a booth there and very exciting topics in the area of decision intelligence, how to build your decisions, how to get them into production, and also how to utilize them in an agentic setup. Diana, thank you so much for coming back on the show and telling us more about SAS Intelligent Decisioning. Thank you, Diana, for the scoop on Intelligent Decisioning. As she said, if you want to learn more about Intelligent Decisioning in person, SAS Innovate is the place to be. And we will be there for our next show. SAS Innovate 2025 is our premier conference for business leaders, technical users, and SAS partners.

We'll have inspiring keynote speakers like Brene Brown, Alfonso Ribeiro, and DJ Jazzy Jeff, some practical hands-on workshops, insightful breakout sessions, and much, much more. It's all happening from May 6 through May 9 in Orlando, Florida. You can visit sas.com slash innovate to learn more and to register.

I'll be there, and I hope to see you there, too. Well, we've reached the end of our show. If you're watching this on YouTube, why not give us a like, subscribe to the SAS Users YouTube channel, and click that bell so you'll get notified for new videos and when we go live for our next show in May. I've been your host, Tiago De Souza. Thanks for watching, and I'll see you next month at SAS Innovate.

2025-04-21 02:02

Show Video

Other news

Getting Data Off a Failed Pre-Built NAS 2025-05-04 02:20
Exploring Graph Analytics in Electromagnetic Spectrum Management - Webinar with Expression Networks 2025-05-02 08:11
Радиометр РК-01 2025-04-30 23:55