Watch this Ask the Expert session to discover how Intelligent Decisioning on SAS Viya can help streamline credit originations, improve risk management and boost productivity.
Watch the Webinar
You will learn:
- How some of the new features can be applied to credit risk decisioning.
- Best practices from SAS credit risk decisioning experts.
The questions from the Q&A segment held at the end of the webinar are listed below and the slides from the webinar are attached.
Q&A
If I make a change to a part of the decision (e.g., a rule or data variable), can I see what impact that has on other decisions? Is there any form of lineage?
Yes, in all areas of the decision flow—whether within specific objects or the list of available variables—you can use an option called “use by” or “lineage.” This shows where an object, such as a rule set or data variable, is used either within the same decision flow or across other flows. This helps you understand the impact of changes.
Is there any way to test bulk data not from the database but using manually designed test cases to cover all scenarios that could happen in production?
Yes. While today’s demo tested against a database, the Scenarios tab allows you to test specific cases. You can upload scenarios via Excel or input values manually. This lets you compare different scenarios across decision versions and trace how variables are generated through the flow to the final output—helping you understand what’s happening behind the scenes.
You covered strategy testing and monitoring, but how are the models in the decision flows monitored?
Monitoring models is essential because they deteriorate over time due to population shifts. Feedback from onboarding strategies flows back to the centralized model repository, generating performance metrics like accuracy, drift, and fairness. Risk models also have predefined KPIs, such as those in Basel Working Paper 14, and users can add their own KPIs for automated documentation and monitoring.
Can Intelligent Decisioning be called or integrated with external systems such as another Loan Origination system or a CRM system?
A. Yes. Intelligent Decisioning can be triggered via an API, batch ETL process, or scheduled job. For systems like loan origination or CRM, it typically involves an API call to Intelligent Decisioning, which responds in real time with decisions and outcomes.
Where are decisions and strategies published to?
Decisions can be published to either a batch-based engine or a real-time engine. The real-time engine uses SAS Container Runtime (SCR), a lightweight container environment that scales horizontally and doesn’t require Viya. It can handle over 20,000 transactions per second. By Q3, SAS Container Runtime will also support batch processing, offering a lightweight option for that as well.
Are there governance features to reduce strategy change risks like check-in/check-out and version rollback?
Yes. Users can check out objects, make changes locally, and commit them back to the master version. This supports auditability and traceability. Version tagging helps with release management, allowing you to tag objects for specific releases. You can also compare versions side-by-side—either code or objects. Workflow Manager enables customized governance processes, assigning tasks for review, scoring, testing, or documentation. A task window will soon allow designers and reviewers to manage these tasks directly within Intelligent Decisioning.
Can this help integrate models and decisioning with Salesforce?
Yes, an upstream system can call SAS Intelligent Decisioning using an API in real time. Alternatively, a batch process can be triggered using an ETL process or a scheduled job where SAS Intelligent Decisioning pulls data from a direct DB integration.
The examples focused on consumer credit (credit cards/personal loans). Could this process be used for commercial clients as well?
Yes, we have a number of financial service providers using SAS Intelligent Decisioning for both consumer and commercial decisioning. For commercial lending, it is possible to automate the lending process by designing strategies compatible with the industry characteristics and credit policy requirements. Strategies can include segmentation, ratio calculation and analysis, hybrid, judgement-based or statistical scorecards including qualitative and quantitative variables, risk-based pricing, conditions and guarantees and automated credit file analysis through use of LLM’s. Segmentation trees also allow streamlining the approval process across the different committees and approval levels. SAS Intelligent Decisioning can also easily integrate with a case management tool such as SAS Visual Investigator or a third-party tool improving the user experience.
SAS Intelligent Decisioning is also used across multiple use cases beyond Credit Risk Decisioning, such as Fraud (e.g., Payments Fraud), Compliance (e.g., KYC onboarding), and Marketing.
Can this decision engine be called from another LO system via WS or API?
Yes, an upstream system can call SAS Intelligent Decisioning using an API in real time. Alternatively, a batch process can be triggered using an ETL process or a scheduled job where SAS Intelligent Decisioning pulls data from a direct DB integration.
Can you provide information about testing the full flow as standalone, not from the LO system.
Carla covers testing during the demo. You can run tests at various levels of the decision flow—either at the full flow level or at the object level (segmentation trees and rulesets).
After changing a rule or updating a component in the Design environment, can we test before overriding the production version.
Yes, the standard architecture is to design and test strategies in a Design environment before promoting to Production. Depending on the change, it may also be tested in a Test/Preprod environment. Carla will touches on that process at the end of the demo.
If I need to change a rule, can I know which other objects it’s used in to evaluate the impact of changes?
Yes, there are various lineage features that allow you to see where objects or data variables are used—either within the same decision flow or across other flows. We may cover this live as well.
Can different users handle Build, Test, and Publish tasks?
Yes, Carla covers this. Workflow Manager allows you to customize a governance process that assigns tasks to different users.
Can all scorecards integrated in the flow also have separate monitoring reports and be shown on a dashboard?
a. Yes, both SAS Model Manager and SAS Visual Analytics are fully integrated, allowing you to monitor the performance of models and decisions in production or from tests.
Is this tool designed to work within the SAS Viya environment?
Yes, Intelligent Decisioning is part of the Viya environment and is fully integrated with other tools in Viya.
Does SAS offer a course on this?
Yes, there are two courses on the SAS Education website that cover the essentials and beyond.
Can we have a matrix-style setup in Flow to configure LTV % for secured products?
There are different ways to configure values based on 2 or more dimensions. In segmentation trees, for instance, a matrix can be built to account for product type and the risk segment defining LTV’s across the two dimensions. It is possible to add additional segmentation layers to the matrix for a more granular LTV allocation. Advanced lists, are another feature that allows allocating values based on multiple keys in a table format allowing to define LTV based on multiple conditions. With this feature, it becomes easily possible to update the LTV values on the fly without having to alter the decisioning strategy that is in production. A third method consists of having a table in an external database that contains this information while the decision can look up the LTV values on the fly.
Recommended Resources
Modernizing Risk Modeling and Decisioning: Practical Strategies for Analysts and Managers
How To Accelerate Governed Decision Delivery With SAS ID Workflows
Be Future-Ready with an integrated approach to Enterprise Customer Decisioning
Please see additional resources in the attached slide deck.
Want more tips? Be sure to subscribe to the Ask the Expert board to receive follow up Q&A, slides and recordings from other SAS Ask the Expert webinars.
