A demonstration of SAS Viya Copilot in Model Studio, now available on the latest 2025.12 Release.
Watch video
What you will learn
Learn how SAS Viya Copilot for SAS Model Studio acts as an intelligent assistant to streamline and enhance the process of building machine learning models. Here's some of what Copilot can do for you:
Build and Refine Machine Learning Pipelines: See how Copilot helps you construct predictive models (e.g., for loan default risk) by suggesting next steps, adding nodes, and dynamically generating robust modeling pipelines.
Explore and Optimize Models: Experience the exploratory process where Copilot tries multiple model combinations and settings, updating you on improvements and fit statistics (like KS statistic) to help select the best modeling approach.
Automate Model Selection and Tuning: Watch Copilot automatically add recommended nodes (such as imputation and gradient boosting), set optimal parameters, and explain each setting. Learn how Copilot can run models, summarize results, and provide high-level overviews, key findings, feature importance, ROC reports, and fit statistics.
Interpret and Compare Models: Discover how Copilot explains model results, identifies potential issues (like overfitting), and suggests improvements. See how you can add challenger models (e.g., logistic regression), chain actions, and run improvement processes to optimize model performance.
Understand Model Differences and Use Cases: Learn how Copilot describes the strengths and use cases of different model types (gradient boosting vs. logistic regression), providing links to SAS documentation for deeper exploration.
Compare and Register Champion Models: Follow the process of running model comparisons, reviewing insights, and registering the final champion model into Model Manager for deployment—enabling use by AI agents in real-world applications like loan approvals.
Get On-Demand Help and Explanations: See how Copilot answers questions about machine learning concepts and SAS Model Studio features, supporting both novice and experienced data scientists.
... View more