Model cards, A/B test nodes, glossary REST-API & more | SAS Viya July 2024 Release
NEWS YOU CAN USE
In this month’s edition of our SAS Viya email series: Model cards are here in SAS Model Studio, you can try out the A/B test nodes in SAS Intelligent Decisioning, we’ve released a REST-API for the glossary inside SAS Information Governance and four new steps have been added to SAS Studio. Plus, learn how to become a more efficient developer with SAS Viya in David’s tip of the month.
NEW VIYA FEATURES
Check out these new features and what they mean for you.
Product: SAS Model Manager
New feature: model cards
Description: Just like a nutrition label for your model, a model card details an AI model’s training data, development process, accuracy, model drift, related fairness assessments and governance details. This transparency helps identify the model’s strengths and limitations, promoting trust and encouraging responsible and ethical use by all stakeholders. Learn more.
Product: SAS Intelligent Decisioning
New feature: A/B test nodes
Description: You can use an A/B test node to direct input records into a champion path or a challenger path for testing. Each path includes one or more objects such as rule sets, treatment groups, models or other objects. Learn more.
Product: SAS Information Governance
New feature: glossary REST-API
Description: The glossary inside of SAS Information Catalog now has a REST-API that can be leveraged to programmatically interact with both Terms and Term Types. Learn more.
Product: SAS Studio
New feature: four new steps
Description: Four new steps have been added, two in Machine Learning (Automated Feature Engineering step and Support Vector Machine step) and two in Optimization & Network Analysis (Cycle Detection step and Reach Network step). Learn more.
SPOTLIGHT USE CASE
Model cards
One practical way to put model cards to work is using them to understand the specific problem the model is designed to address, its limitations and out-of-scope use cases. For example, if a model exists and you’re wondering if you can use it on data from children, you can consult the model usage section and see that the model should not be applied to patients younger than 26. This helps product managers, developers and executives make informed decisions about which models to deploy for specific applications, understand potential risks and ensure responsible AI practices. Learn more.
TIP OF THE MONTH
Brought to you by David Weik
