SAS Viya Copilot Explained: Building Machine Learning Pipelines in Minutes, Not Hours
Recent Library Articles
SAS Viya Copilot is a set of software as a service (SaaS) features and capabilities that use large language models (LLMs) to provide users with a more intuitive and accessible way to work with SAS Viya offerings. SAS Viya Copilot is for developers, data scientists, citizen data scientists, and business analysts who are writing code, analyzing data, building machine learning model pipelines, and doing more across the data and AI life cycle.
Already a SAS Viya user? Then you know you’ve got a wealth of capabilities at your fingertips. We’ll explore those together, giving you access to all the insights you’ll need. Join us on December 10 at 11 a.m. ET.
I open SAS.
I submit a macro (just submit it, so nothing happens except it puts the macro into memory). A line in the macro points to a formats library.
I call the macro, so the guts of the macro runs.
When the macro runs it creates a dataset. After running the macro I do a PROC FREQ on that dataset. All the formats are applied correctly.
Then I call the macro again. I do not submit the macro itself again, rather I just call (run) it. I change nothing and I do nothing else. All I do is call the macro a second time.
Then I do that same PROC FREQ mentioned above and this time some...not all, but some...of the formats are not applied.
So the 1st time I run the PROC FREQ I see all words (the formatted values) instead of their underlying numbers.
And the 2nd time I run the PROC FREQ I still see mostly words (the formatted values), but for some of them I see their underlying numbers instead.
And every subsequent time I run it I see the mostly words but some underlying numbers instead.
IOW, it works properly after I open a new SAS session and run it for the first time. And it doesn't work properly every subsequent time I run it.
Any idea?
... View more
What if you could combine the flexibility of the cloud with the performance of your on-premises infrastructure? In a joint validation, Intel, Red Hat and SAS explored how SAS Viya runs on Red Hat OpenShift Virtualization with Intel Xeon 6 processors. From proactive autoscaling with SAS Workload Orchestrator to live migration of running workloads, discover how cloud-like elasticity and enterprise analytics performance can come together on-premises.
... View more
SAS VIYA 4 | FINANCIAL CRIMES
SAS Viya Copilot Prompt Library for Financial Crimes
Reusable prompts for governed data, analytical modeling, investigations and fraud decisions.
Jump to a section
SAS Data Governance
SAS Data and AI Studio
SAS Model Studio
SAS Model Manager
SAS Visual Analytics
SAS Visual Investigator
SAS Fraud Decisioning
BEFORE YOU START
Adapt the examples. Replace table names, variables, thresholds, projects and terminology with values from your SAS environment. Review outputs. Validate generated code, recommendations, summaries and rules before operational use.
GOOD TO KNOW
These prompts are starting points for qualified SAS users working with authorized data and projects.
Capabilities and available actions can vary by product configuration and permissions.
SAS Viya Copilot in SAS Data Governance Create a trusted financial-crime data foundation Back to top ↑
DATA DISCOVERY
USE CASE
PROMPT
EXPECTED OUTPUT
Find customer data
“Find governed data assets containing customer and account information for a financial-crime analysis.”
Lists potentially relevant governed data assets.
Find transaction data
“Find data assets that contain transaction amount, date and time, channel and counterparty information.”
Lists potentially relevant governed data assets.
Find risk data
“Find data assets that include geography, country and customer risk-indicator information.”
Lists potentially relevant governed data assets.
Find payment data
“Find data assets related to counterparties, beneficiaries and payment channels.”
Lists potentially relevant governed data assets.
Find analysis data
“Find data assets I can use to detect unusual transaction behavior across accounts.”
Lists potentially relevant governed data assets.
SAS Viya Copilot in SAS Data and AI Studio Prepare and explore the data Back to top ↑
DATA PREPARATION
USE CASE
PROMPT
EXPECTED OUTPUT
Get product help
“Explain how to remove duplicate rows from a table in SAS Data and AI Studio.”
Explains the process using SAS documentation.
Profile with code
“Generate Python code to identify missing and duplicate records in a dataset.”
Generates Python code for review and execution.
Join source tables
“Generate SAS code to join tables [table X], [table Y] and [table Z] and keep the amount, date and time, channel, geography and counterparty variables.”
Generates adaptable SAS code for joining the tables.
Engineer features
“Generate SAS code to calculate transaction count and average transaction amount per account for each 30-day period.”
Generates SAS code for behavioral features.
Explain code flows
“Explain what this code or flow does.”
Explains the code or flow in plain language.
SAS Viya Copilot in SAS Model Studio Detect unusual behavior Back to top ↑
MODEL DEVELOPMENT
USE CASE
PROMPT
EXPECTED OUTPUT
Select an approach
“I want to identify customers or accounts whose activity differs from their normal behavior. What anomaly-detection or machine-learning approaches should I consider?”
Recommends modeling approaches and selection considerations.
Build a pipeline
“Build a pipeline to identify unusual account behavior in this data.”
Builds a pipeline with relevant preprocessing and modeling nodes.
Choose model inputs
“Which variables would you recommend as inputs for detecting anomalies, and why?”
Recommends inputs and preparation considerations.
Compare models
“Compare the models in this pipeline and explain which is most effective at identifying unusual behavior.”
Compares results using relevant assessment measures.
Interpret activity
“Explain which variables contribute most to the model’s identification of unusual behavior and how they may help an investigator understand the result.”
Explains influential variables and model behavior.
SAS Viya Copilot in SAS Model Manager Govern and operationalize models Back to top ↑
MODEL LIFECYCLE
USE CASE
PROMPT
EXPECTED OUTPUT
Understand project
“Summarize this financial-crime detection project, including models, versions, champion model, available performance information and publishing status.”
Summarizes the project and its registered models.
Compare candidates
“Compare the models in this project and explain the differences between them.”
Explains differences among registered model candidates.
Manage champion
“Set [model name] as the champion model for [detection project].”
Applies the designation using supported confirmation steps.
Monitor performance
“Summarize the latest available performance-monitoring results. Highlight material changes in performance or input data that require review.”
Highlights available performance and data changes.
Publish approved model
“Publish the approved champion model to [publishing destination] and report the resulting status.”
Publishes the model and confirms its resulting status.
SAS Viya Copilot in SAS Visual Analytics Understand patterns and hotspots Back to top ↑
VISUAL ANALYSIS
USE CASE
PROMPT
EXPECTED OUTPUT
Explore suspicious activity
“Analyze the detection results and identify notable patterns by transaction type, customer segment, geography, channel and time.”
Summarizes notable patterns and areas for analysis.
Create hotspot report
“Create a report showing where anomalous activity is concentrated by geography, transaction type and customer segment. Include filters for date, channel and risk level.”
Creates a report with relevant visualizations and controls.
Examine trends
“Show how the volume and value of anomalous transactions change over time. Break results down by channel and transaction type.”
Creates time-series visualizations and summarizes changes.
Compare segments
“Compare anomalous activity across customer segments using flagged transactions, total value and available risk measures.”
Compares segments and explains notable differences.
Explain a report
“Explain the key insights in this report, including concentrations of anomalous behavior and patterns warranting further investigation.”
Explains the report’s principal findings.
SAS Viya Copilot in SAS Visual Investigator Investigate connected activity Back to top ↑
INVESTIGATION ASSISTANCE
USE CASE
PROMPT
EXPECTED OUTPUT
Explore the network
“Show the customers, accounts, counterparties and transactions connected to this account. Highlight shared attributes and repeated relationships.”
Shows connected entities and their relationships.
Prioritize connections
“Which connections in this network warrant closer review based on transaction frequency, value, timing, shared attributes or proximity to other alerted entities?”
Prioritizes connections using available investigation information.
Summarize findings
“Summarize the key entities, transactions and relationships identified in this investigation. Distinguish observed facts from analytical indicators that require validation.”
Separates established evidence from indicators requiring review.
Identify next steps
“Based on the information currently available, what entities, relationships or transactions should an investigator review next, and why?”
Recommends follow-up areas tied to available evidence.
Retrieve object details
“Show [customer name]’s date of birth from the object record.”
Returns the authorized field value when available.
Navigate records
▸ Switch to [customer name].
Opens the selected object or tab.
Search parties
“Show me any people with the surname [surname].”
Lists objects matching the search criteria.
Locate a party
“Do we have any information on [customer name] from [location]?”
Returns matching records or reports no matches.
Find linked claims
“Show any open claims associated with this object or its linked objects.”
Summarizes related open claims using available relationships.
Get product help
“Explain how to create a relationship between objects in SAS Visual Investigator.”
Explains the process using product documentation.
SAS Viya Copilot in SAS Fraud Decisioning Uncover fraud with natural-language rules Back to top ↑
RULE AUTHORING
USE CASE
PROMPT
EXPECTED OUTPUT
Understand available data
“What is the current message schema for this rule?”
Identifies the available schema and profile variables.
Understand a variable
“What is “minutes since last mobile event,” and how can I use it in a fraud rule?”
Explains the variable and its potential use in rule logic.
Generate fraud rule
“Can you help me write a rule? If the message is an ATM withdrawal that is not being made in the cardholder’s home country and the amount >400, then decline the message and generate an alert.”
Generates DS2 rule code using available variables.
Explain fraud rule
“Explain what this code means.”
Explains the rule’s conditions and resulting action.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
... View more
anyone knows the one below?
ERROR: Windows error code: 21 in hc_disk_normal_read for
E:\sastemp\_TD16080_LAPTOP-LF4NQTV2_\sastmp-000000005.sas7butl
Work with pc/sas, ever session to restart sas, get this and need reboot sas.
... View more
I am using SAS Viya 4, I have about 50 .sas programs in a folder on SAS Server. I would like somehow to find which of these programs have specific text, such as "Loans Approved" without the quotes.
How can I do this?
... View more