Skip to content

Release Notes V1.11.0

Version 1.11.0
Release Date 07th May 2020
Changes
  • Ability to use Runtime Parameters to allow providing parameters during the artifact runtime
  • Export documentation for models registered in Model Studio
  • Add ability to provide weights to be used in reports during simulation time for Classification models and some reports in Regression models
  • Enable the file-upload and manual definition sources by default in all definition boxes
  • Allow Draft items to be accessed in Notebook through the corridor package
  • Create and allow the artifact to be downloaded through the UI - in every module
  • Roll out Batch Execution to all modules
  • Expose the queue routing for Celery Spark Workers using a CeleryQueueHandler
  • Expose the username of the user when reading data from the data lake
  • Unify the DataSourceHandler definition, so that it is defined only on the API side and is sent to the celery workers.
  • Improve speed to load input boxes and dropdowns with a large number of items
  • Use the local time zone of the user instead of showing the UTC time zone for timestamps
  • Disable snake-case formatting for aliases by default
  • The “Advanced Settings” tab has been split into “Settings” and “Resources”
  • Improved error messages when saving a product, framework, policy, etc. duplicate name or alias
  • Mention the threshold used for calculating model metrics with threshold
  • In the model classification reports for KS calculation - show the max separation in the decile table’s title and stop showing the KS based on thresholds.
  • During model registry, simplify the Train/Validation dataset specification. Also, add the ability to label the locations provided for the datasets.
  • Various performance enhancements in the artifact
Fixes & Issues
  • Show histograms (for QualityProfile, DataElement, Feature, and Model simulations) correctly for smaller data and data with outliers
  • Handle issues in some model-reports when the model’s entity and dependent’s entity are different
  • Handle dependency issues with PMML and numpy in CentOS7
  • Fix issues when creating a feature with product-type-configs or global-variables
  • Fix issue when running a simulation with only Additional Filters
  • Show scroll bars correctly for Plotly charts in Windows
  • In dataset, show accurate errors if the Input Distribution report fails
  • Suppress the passlib warnings when aliases with capitals were used
  • Fix issue when replacing
Upcoming Features
  • Function Registry
  • Add the ability for users to write aggregates with lists
  • Enhanced Authorization
  • ID-based filters
  • Enhanced SAML
  • OWASP-Security checks
  • Custom Dashboard display