Welcome to the latest One Model product release update. This article provides an overview of the product innovations and improvements to be delivered on 22 November 2023.
- Building on our last release - 2023.11.08 where we released an enhancement for Single Value Tiles that allowed users to change the color of the value displayed, we have now added options to customize:
- Text style - regular, bold, italic, and bold italic,
- Text size - input box for custom sizing, and;
- The option to ignore the decimal value.
(ref 16892, 17892)
- We made a change to the visibility of the filter bar on the Home Page and Storyboards, as follows:
- Home Page - There is an Application Access Role permission called “CanChangeHomePageFilterSet” that allows a user in a given role to be able to add or change filters on the Home Page. Before now if there were no default filters on the Home Page then a user would see an empty filter bar. As of this release, in this situation the filter bar won’t be visible to the user.
Storyboards - Access to change filters on Storyboards is a little different, this is controlled as part of Storyboard Sharing. When sharing a Storyboard there is an option to allow particular roles to “CanChangeFilterSet”. Similar to the scenario above, if a user is in a role that does not allow them to change the filter set on a particular Storyboard, AND that Storyboard doesn’t have any default filters then the filter bar won’t be visible to the user.
Data Processing Improvements
- We've optimized our Processing Script Validation step to now take up to 10x less time than it used to! This was done by calculating the data types for columns in each Processing Script object once, rather than calculating them each time they are referenced - both in the current object, and any dependant objects. (ref 17608)
- We separated Machine Learning Models and Data Augmentations to facilitate the creation of regression Recipes. We also simplified the workflow so it is more intuitive. (ref 2399)
- The single ‘Add Augmentation’ button at the top of the Augmentations page has been replaced by two buttons:
- ‘Add Machine Learning Model’
- ‘Add Data Augmentation’
- When ‘Add Machine Learning Model’ is selected, the first step after assigning a name is to jump into a ‘Recipe’ or select a ‘Destination’, instead of ‘Augmentation Type’ as you did previously.
- We have replaced the term ‘Augmentation Query’ with ‘Recipe’.
- If ‘Recipe’ is selected and ‘Custom / Advanced Model’ is selected as the Recipe Type, you will need to define the Problem Type the model is intended to address.
- Selecting ‘Regression’ as the Problem Type will hide the option to configure target overrides in the recipes as regression models predict continuous rather than categorical targets.
- The workflows for configuring Machine Learning Models from Data Destinations and Data Augmentations has not changed significantly.
Minor Improvements & Bugs Squashed
- A customer reported an issue with their Role Based Security (RBS) involving a contextual role that uses dimension “Person (used for RBS)” which is a Slowly Changing Dimension. If the data access role is set to “theirs or one of their descendents”, the correct results are returned. But, then if the role is adjusted to be “not theirs but their descendents”, the results are incorrect as some intersections were dropped resulting in not all of the descendents being returned which would affect metrics like Headcount. We made a change to the way we handle data access roles and contextual security rules involving dimensions that are set to ‘not theirs but their descendents,’ and ‘theirs or a sibling of theirs’ and ‘one of their descendants’, so that all of the intersections are properly generated and the correct results are being returned. (ref 17778)
- We fixed a bug that prevented users from saving an Org Chart if you had the same metric in the data item field and the details panel. (ref 17586)
- We fixed a bug where the Home Page filter labels were not showing in the tiles on the Home Page. (ref 15476)