Financial Services Model Management and Governance (OFS MMG) Application Pack, v8.1.2.0.0 Maintenance Level Release #5 (8.1.2.5.0)

Description

OFS MMG 8.1.2.0.0 Maintenance Release #5 8.1.2.5.0 (ID 35744607).

This release of the OFS MMG Application Pack is cumulative of all enhancements and bug fixes completed since the v8.1.2.0.0 release.

NOTE: For information about the Hardware and Software Requirements for this release, see the Oracle Financial Services Analytical Applications 8.1.2.0.0 Technology Matrix.

How to Apply this Release?

For instructions on how to install this OFS MMG Application Pack Release, see the Oracle Financial Services Model Management and Governance Application Pack Installation Guide, Release 8.1.2.5.0.

New Features and Enhancements in this Release

1 Workspaces

� Workspace Population: Populate a subset of selected tables

� Simple Expression Builder provided for SQL Filters in the Workspace Population screen

� Added escape-delimiter-character parameter in File Data Store

� Data Population filters support custom and system runtime parameters

� Ability to use Import Archive and Templates to update Data Source schema without changing the selected database objects

2 UI/UX Enhancements

� Sidebar menu replaced with the all new Mega menu

� Data Studio menu options moved out from workspaces to the global menu

� Improved UI messages and logging

� Standardized layouts and controls on all screens

� UI/UX improvements in line with Redwood design

3 Model Pipelines

� New split Pipeline Builder UI for easier interaction with canvas and node configurations simultaneously

� New widgets for Feature Engineering, Model Training, and Stream Analytics

� Ability to create sub-pipeline containers

� Stitch already published Model Pipelines into a composite Pipeline

� Support for all kinds of dynamic forms like Data Picker and Sliders, in Parameter Sets and executions UIs

� Automatic cleanup of �dummy node� or non-existence paragraphs in Pipeline

� Support of Bulk edit of Parameters Set values

� You can now customize number of approval and escalation levels

� Integration of Model Deployment workflow with Notifications

� Ability to view Pipeline or Notebook by default when a Model is opened in Sandbox. This is via a service level configuration.

� Breadcrumbs added to the Pipeline Details screen

� Switch to different versions within the Canvas screen

� API Registry access allowed from Model Pipelines screen

� Introduction of new screen for the Deployment Workflow

� Option to show Executed Only paragraphs introduced in Outputs/Compare of Pipeline Execution History

� Option to attach Conda and Parameter Set dependencies from the Publish model screen

� When adding a new paragraph, application automatically inherit the previous/ incoming paragraph's link type.

4 Identity Management

� JIT sync of username/email ID

5 Graphs

� Out-of-the-box graphs analysis notebooks supported. Additionally, you can now create and manage notebooks for analyzing graphs directly from the Graphs details screen.

� PGX upgraded to 23.4.3 version

� Enhanced Data pipeline interface

� Reset Graph metadata utility resets all metadata related to a particular graph ID

6 Model Catalog

� Model builder optionally generates model evaluation metrics

� Model builder supports scoring of external model

� Ability to upload Model to the server from the client system

� Access Model Catalog from Model Pipeline Details screen

� Model Technique/Library � Support to Multi-line for all the Signatures

7 Kafka Support

� Automatic discovery and listing of Topics in the registered Kafka clusters

� New overview screen to see sample records and other details of a Topic

8 Datasets

� New overview screen to view features, profile metrics, and sample records at a glance

� New data visualization and feature maintenance screen introduced

� Ability to download sample records of datasets in CSV format

� Create New Feature option to import any custom Python library introduced

� One-click Data Drift configuration by pre-filling all necessary fields enabled

� Support for attaching Conda environment during Dataset creation

� Ability to update Data elements with different Data sources and their database objects provided

9 Data Studio

� Data Studio upgraded to 23.4.3 version

10 Data Pipeline

� Data pipelines are now available as Standalone. With this enhancement, you can:

o Define a Standalone pipeline to move data from one data source to the configured/mapped target data source.

o Create and manage data pipelines for efficient data flow and processing.

To execute this pipeline, you must create a batch/task with component as DATA_PIPELINE in the Scheduler Service component.

For more details, see MMG User Guide.

11 General

� By default, a Sample Workspace has been added which provides a demo view of the MMG concepts/features.

For more details, see MMG User Guide.

Known Issues and Limitations in this release

Limitations:

� Encode categorical feature transform fails with MODIN data frame library if the selected features contain different values.

� Cache functionality might not work as intended while using MODIN data frame library.

� Select all data sourcing objects does not work as intended during workspace provisioning.

� Error log table creation fails if the column data types are LONG, CLOB, BLOB, BFILE, and ADT during Sandbox data population.

� The PDF of the Model Report does not contain data in the output section.

� Unable to perform dataset cache with Modin library.

� In the Log in screen, the user name provided must be in uppercase.

� Miniconda required for conda provisioning must be installed on User Home Directory.

� Paragraph names if added/modified from Notebook tab will not be persisted.

� Deployment of Models published from the Model Summary screen will not promote the associated dependencies such as Graphs, Parameter Sets, Datasets, and Models.

NOTE: Publishing of Models option from the Model Summary screen will not be deprecated and continue in the upcoming releases.