Chiron App 2.0 & 2.1 release: Enhanced Object Management and Simplified Data Handling

We are excited to announce the release of Chiron App 2.0 to new and existing clients. This latest update introduces new features and improvements to streamline your work within the application. What’s new and what can you expect with the 2.1 release?

• Flexibility in viewing: Toggle between the default object card view and the list view easily. This feature enables you to see more objects on a single page in a paginated and tabular format. 
• Extended sorting attributes: Sorting capabilities within Chiron App 2.0 have been expanded. Now you can navigate and organize your objects easily by ‘last run date’, ‘number of sessions’, ‘specification name’, etc.
• Total object count display:  Have complete visibility on the number of objects you are working on with the new total object count display. 

We have reduced the types and amounts of objects within Chiron App for a more straightforward experience. To this extent, artifacts have replaced datasets and only one type of artifact remains, eliminating the need for data model mapping and simplifying the structure.

Validation checks on datasets (e.g. number of columns, name, type) are also now more transparent, with instance runs providing clarity on data quality issues.

The Python API has been re-designed to make it more user-friendly and readable. With a more abstracted approach to the underlying infrastructure, integration with other technologies is now more seamless.

• Permanent Deletion of Artifact Versions: In the upcoming Chiron App 2.1 release, users will be able to permanently delete artifact versions (e.g. bad data and/or old data), aiding in better data management and data retention policy.
• Batch Deletion: The new version will also introduce the ability to delete multiple files at once, further simplifying data housekeeping.

We offer the option to pre-install best practice based model risk management use cases upon request. These include analyses such as data quality checks and validation of binary target models. Additionally, these objects can be tagged  for sharing and usage by others, while avoiding that they’re unnecessarily modified and their full functionality is preserved. Our intent is to provide model risk stakeholders with ready-to-use quantitative analyses to support their work from the go-live.