Welcome to the 2020.05.13 product release. This article provides an overview of the product innovations and improvements to be delivered on 13 May, 2020. The article is structured as follows:
User Experience
One Ai Innovations
Performance & Platform Improvements
-
Improvements to Data Pipeline Processing
-
Minor Improvements and Bugs Fixed
-
One Ai Bugs Fixed
User Experience
-
In the initial delivery of the new User Experience we focused on all of the core user capabilities and one of the last features to fully replace the functionality of the traditional Dashboards is embedding. In this release we have added the ability for customers to embed Storyboards or the entire One Model application into another site/portal etc. The existing administrative settings for setting up embedding in the company Admin screen all carry over, but the embedding links are new, so if you have Dashboards embedded you will have to update your embedding links when adopting the new UI. (ref 4316)
One AI Innovations
-
Metadata Consumption: In this release we’ve made the first of several upcoming updates that allow One AI to use ‘context’ gained from data relationships within the core One Model application to inform data preparation decisions. Data shared with One AI will now contain a metadata file, when available, which lets One AI know which tables and what relationships exist for any given data point. This type of data opens up a lot of exciting functionality. One AI can now use metadata to understand difficult to detect collinearity between variables to make more intelligent dimensionality reduction and feature selection decisions, resulting in more stable predictive models. (ref 781)
-
Helpful Forecaster Errors: We’ve been gradually adding advanced forecasting functionality over the last several releases. Historically, the error messages provided by the forecaster didn’t provide enough information to allow users to easily resolve problems. In this release we added the first set of ‘prescriptive forecaster errors’. (ref 824)
Performance & Platform Improvements
Improvements to Data Pipeline Processing
-
In this release we delivered a new feature within the Data Pipeline Processing editor that shows whether another user has the same processing script open to alert users so that they can avoid any issues with version control. (ref 3174)
Minor Improvements and Bugs Fixed
We fixed a bunch of bugs in the display and branding capability as part of the new User Experience, including:
-
We made a lot of improvements under the hood of the new Storyboard designer in this release. One example you will notice is that when you enter or exit Storyboard edit mode your charts and tables aren't being re-queried which improves the overall speed and feel of the user experience. (ref 4337)
-
Fixed an issue where a narrow text tile with a title only would show a scrollbar even if there was no text in the body of the tile. (ref 4304)
-
Fixed an issue where the page background wouldn’t flow across the whole page depending on the browser zoom level and length of the page. (ref 4592)
-
Fixed an issue when making selections within a dimension in the Storyboard filter bar the app sometimes switches focus and opens the add new filter option. (ref 4498)
-
Fixed an issue where the tile specific menu wouldn't display correctly for very narrow tiles. (ref 4280)
-
Fixed the default page background colour to #F5F5F5 (White Smoke for those playing along) (ref 3887)
-
Allowed an Admin user to select the colour to be used as the page and Storyboard background (i.e. allowing you to change from White Smoke to White, or to #FF69B4 - as always sometimes less is more when it comes to changing the Branding settings :). (ref 3888 & 3891)
-
Corrected the colour applied to the toggle buttons in the Storyboard Settings menu that enable or disable filter labels being displayed on tiles. (ref 4484)
-
Corrected the action button colour in the session timeout pop-up (ref 4541)
One AI Bugs Fixed
-
Bias detection and shift analysis bug: In the previous release we added the ability to calculate group fairness metrics to try and help users analyze bias datasets. If a column that was needed for the prescribed group fairness metric was dropped from the data set due to shift analysis One AI would crash. This behavior has been fixed in the latest release.(ref 822)
Comments
0 comments
Please sign in to leave a comment.