What's new and changed in Watson OpenScale

Watson OpenScale updates can include new features, bug fixes, and security updates. Updates are listed in reverse chronological order so that the latest release is at the beginning of the topic.

You can see a list of the new features for the platform and all of the services at What's new in IBM® Cloud Pak for Data.

Refresh 9 of Cloud Pak for Data Version 4.0

A new version of Watson OpenScale was released as part of Cloud Pak for Data Version 4.0.9.

Operand version: 4.0.9

Security fixes
This release includes fixes for the following security issues:
  • CVE-2020-10663
  • CVE-2022-24785
  • CVE-2021-43138
  • CVE-2021-37136
  • CVE-2021-37137
  • CVE-2021-21409
  • CVE-2021-21295

Refresh 8 of Cloud Pak for Data Version 4.0

A new version of Watson OpenScale was released as part of Cloud Pak for Data Version 4.0.8.

Operand version: 4.0.8

Security fixes
This release includes fixes for the following security issues:
  • CVE-2021-44906

Refresh 7 of Cloud Pak for Data Version 4.0

A new version of Watson OpenScale was released as part of Cloud Pak for Data Version 4.0.7.

Operand version: 4.0.7

This release includes the following changes:
New features

The 4.0.7 release of Watson OpenScale includes the following features and updates:

Upload pre-scored test data
Now you can upload test data that was already scored for a model deployment so that Watson OpenScale does not spend resources scoring the test data a second time. Prediction and probability data are uploaded with the test data. For more information, see Manage model risk.
New fairness metrics and explanation algorithms
You can now use Watson OpenScale Python client version 3.0.14 or later to generate new fairness metrics and additional algorithms to explain model transactions. For more information, see Metrics computation using Python SDK.
Security fixes
This release includes fixes for the following security issues:
  • CVE-2019-19919
  • CVE-2021-22569
  • CVE-2021-31684
  • CVE-2018-3745
  • CVE-2015-8315
  • CVE-2021-23369
  • CVE-2021-23383
  • CVE-2019-20920
  • CVE-2019-20922
  • CVE-2020-15250
  • CVE-2021-43797
  • CVE-2022-21699

Refresh 6 of Cloud Pak for Data Version 4.0

A new version of Watson OpenScale was released as part of Cloud Pak for Data Version 4.0.6.

Operand version: 4.0.6

This release includes the following changes:
New features

The 4.0.6 release of Watson OpenScale includes the following features and updates:

Expanded authentication options for custom machine learning engines
In addition to basic authentication, you can now authenticate by using API keys if you have custom machine learning engines.
New usability and performance features
Support has been added through the UI and API for these new features:
  • setting a new database password through the UI
  • defining multiple metric threshold for Quality monitor
  • defining threshold for data accuracy in Drift monitor
  • publishing OpenScale metrics to OpenPages objects of type "deployment"
  • assigning group level access permissions to an OpenScale instance
  • allowing users to specify max_records to process during Quality and Drift evaluation
  • processing large datasets in feedback/quality monitoring
  • configuring a batch deployed python function as a Custom Metric Provider
Security fixes
This release includes fixes for the following security issues:
  • CVE-2021-3807
  • CVE-2021-3918
  • CVE-2021-36090
  • CVE-2021-21290
  • CVE-2021-21295
  • CVE-2021-21409
  • CVE-2021-3426
  • CVE-2019-18348
  • CVE-2019-20907
  • CVE-2021-23336
  • CVE-2018-20852
  • CVE-2021-3426
  • CVE-2022-21704
  • GHSA-gf8q-jrpm-jvxq
  • PRISMA-2021-0037

Refresh 5 of Cloud Pak for Data Version 4.0

A new version of Watson OpenScale was released as part of Cloud Pak for Data Version 4.0.5.

Operand version: 4.0.5

Security fixes
This release includes fixes for the following security issues:
  • CVE-2021-44228
  • CVE-2021-45105
  • CVE-2021-45046
  • CVE-2021-44832

Refresh 4 of Cloud Pak for Data Version 4.0

A new version of Watson OpenScale was released as part of Cloud Pak for Data Version 4.0.4.

Operand version: 4.0.4

This release includes the following changes:
New features

The 4.0.4 release of Watson OpenScale includes the following features and updates:

Support on s390x hardware
You can install Watson OpenScale on Red Hat® OpenShift® Container Platform Version 4.8 clusters running on s390x hardware (IBM Z® and LinuxONE.

Refresh 2 of Cloud Pak for Data Version 4.0

A new version of Watson OpenScale was released as part of Cloud Pak for Data Version 4.0.2.

Operand version: 4.0.2

This release includes the following changes:

New features
The 4.0.2 release of Watson OpenScale includes the following features and updates:
Evaluating custom metrics
You can configure an evaluation of custom metrics to validate your custom monitor thresholds. You can view the results from the Insights dashboard.

Initial release of Cloud Pak for Data Version 4.0

A new version of Watson OpenScale was released as part of Cloud Pak for Data Version 4.0.

Operand version: 1.0.0

This release includes the following changes:

New features
Version 4.0.0 of the Watson OpenScale service includes the following features and updates:
Support for additional capabilities in batch environments
You can now use the following capabilities in batch environments:
  • Generate explanations for every record
  • Detect bias
  • Compute fairness metrics on your training data

For details, see Batch processing.

Support for Db2® as a data source in batch environments
Previously, Db2 was supported as a data source only for online subscriptions. Now, you can use Db2 as a data source in batch environments for payload, feedback, and training data. For details, see Batch processing.
Support for JDBC connections
You can now use a JDBC connection type when you configure data sources. For details, see Configuring the batch processor.
Use a remote instance of Watson Machine Learning
You can now bind a machine learning provider to:
  • Watson Machine Learning on IBM Cloud
  • Watson Machine Learning on another instance of Cloud Pak for Data

In addition, this release of Watson OpenScale provides enhanced usability for local instances of Watson Machine Learning.

Latest version of scikit-learn
Retrain your drift model to use the latest libraries that are available in scikit-learn 0.24. Beginning in Watson OpenScale on Cloud Pak for Data Not applicable, drift detection models that are built on scikit-learn 0.20.1 will stop working. For details, see Configuring the drift detection monitor.