Skip to content

Release Notes V1.22.0

Version 1.22.0
Release Date 21th July 2023

Changes

Data and Model Management

  • Enhanced Experiments to enable use data from external sources.

  • Moved Algorithm from Model Studio to Resources Module making it a generic analytical resource.

Common

  • Snapshots: Enhanced the change log by adding the following snapshot related capabilities:

    • Snapshot Preview: Allows users to view any past snapshot of an object.

    • Snapshot Restore: Undoes recent changes and reverts the object to a previous snapshot.

    • Copy Snapshot: Creates a new object using a copy of an object's past snapshot

  • Monitoring Dashboard: This a new capability that allows users to monitor different aspects of the information and activity on the platform

    • The Activity dashboard tracks user activity (logins, logouts, session times, etc.)

    • The Governance dashboard tracks the governance statuses of all objects (Approval statuses, Reviews statuses, Review due dates, etc.)

    • The Metrics dashboard tracks specific metrics for specific objects (AUC, Accuracy, Drift, etc. for models, Average, Variance, etc. for Data elements,..)

  • Monitoring Alerts: Using the Monitoring Dashboard, Users can set conditions that can trigger alerts if met.

    • Alerts are defined by conditions and severity

    • Actions can be automatically taken if an alert triggers:

      • Send notifications, emails to specific users.

      • Initiate review requests.

  • Custom Fields: Improved capability to enable additional field types, searchability and access control.

  • Report Creation: Enabled easier access to object metadata to simplify report writing.

  • Approval and Review Requests: Allow approval requests or review requests to be sent to anyone in the responsibility without naming a specific reviewer.

  • Error messaging: Provide clearer errors when creating objects with create_data() command in the notebook.

Technical

  • Simplify Ability to run multiple corridor-workers on the same machine by specifying worker name and worker specific configurations.

    • Example use case: Run separate corridor-worker to execute “Fetch Columns”

  • Multiple performance speedups for various APIs.

Backward compatibility

  • Worker queues updated with below changes:

    • apilocal renamed to api

    • celery split into two queues: spark and quick_spark

  • Users without read access to all the precedent items to an object will not be able to export the artifact of the object.