2 New Features and Enhancements in this Release
- Installer
- All port number being used for services are now configurable.
- A new pre-installation utility is now available to validate install configurations like availability of ports, ftpshare/log folders, database connections, etc.
- Data Stores
- Users can now upload new files (.csv and .txt) while registering File based Data Stores.
- Ability to derive the schema information for the File based Data Stores.
- Data Studio Conda Environments
- Users can now register custom/local Conda channels for package installation.
- Workspaces
- Users can now provide database Hints at table-level and SQL Prescripts at schema level for data load performance improvement during Workspace Population.
- Ability to do data repopulation for a subset of tables.
- Datasets
- Introduction of Data Fairness functionality.
- Support of Pyspark option for scaling dataset.
- Model Pipelines
- Introduction of Interactive execution mode in Dashboard for users to pause and review results before providing next set of runtime parameters and continuing execution.
- Support of markups in %python paragraphs to allow users to track outputs and update node type (training/scoring/experimentation) from python script itself.
- New Deployment Overview screen to improved user experience around deployment workflow.
- Addition widgets introduced for feature engineering and model training.
- Model Catalog
- Additional metrics introduced to score/evaluate models built externally and those built from Pipeline/Notebook.
- All models can now optionally be served as a webservice exposing its prediction endpoints.
- Data Studio
- Data Studio upgraded to 23.4.7 version.
- Identity Management
- Option to fetch and provision user groups from OFSAA during user provisioning.
- Scheduler
- Support of providing runtime parameters in ESIC.
- UI/UX Enhancements
- Improved UI messages and logging.
- UI/UX improvements in line with Redwood design.