Release Notes - 2024.04.17

Welcome to the latest One Model product release update. This article provides an overview of the product innovations and improvements to be delivered on 17 April 2024.  

*New* One AI Academy Customer Community Webinar

• The first ever One AI Academy customer community webinar is coming up on Tuesday, April 30th.
  

  Join our Chief Data Scientist, Taylor Clark, and VP of People Analytics Strategy, Richard Rosenow, where they will cover:

  • How generative AI works
  • Learn how One Model is different to other people analytics providers
  • Watch a live demo of One AI Discover
  • Get your One AI questions answered
• Register for the One AI Academy Customer Webinar

User Experience Improvements

Beta - Define Panel

• We have added Geomaps to the Define Panel in Storyboards which enables you to visualize geographic data, as shown below. Current options in the Define Panel include Basic Charts, Lists, Key Values, Sankey Diagrams, Pie Charts, and Free Text Tiles with additional data visualizations still to come. Learn more in our Beta - Define (Explore in Storyboards) Guide(ref 18740)

• Along with the introduction of the Pie Chart visualization to the Define Panel for Storyboards, we have implemented warning and error messages, along with color-coded notification bar indicators to guide you through your query creation. (ref 19342)
  
  
• We've made a small adjustment to ensure consistency in the charting functionality between Explore and Define. In Define, the chart functionality now defaults to "Show Data Labels = On". You can still turn data labels off for each metric in Tile Settings, as shown below. (ref 19463)
  
  

Beta - Custom chart legend

• Now you can use custom chart legend with Pie Charts! Customize your basic chart or pie chart legend by grouping legends by metric name or hiding the legend name all together. This helps decrease chart clutter and redundancies making your legends simpler and easier to read. Learn more in our Beta - Custom Chart Legend Guide (ref 18480)
  
  

Data Improvements

• SuccessFactors Consolidation Enhancement - A feature to batch the consolidation tasks. This is to optimize the consolidation tasks and reduce the risk of a consolidation task getting stuck.
  • Details:

    Batch Consolidation Tasks = unchecked :

• For SuccessFactors extraction, the data is extracted from SF API by batch. The previous design for consolidation tasks consolidates all the batches into a single file. The consolidation process includes building the tables from all batches and merging it all into a single file. This takes quite a while as the process of building the tables takes a huge chunk of infrastructure resources (CPU, memory) on a single Kubernetes pod. This long running task has a higher chance of getting stuck and thus needing to retry the task until it gets completed.

  Batch Consolidation Tasks = checked :

• The new design is to move the table building logic within the batched tasks so the load of creating the tables can be distributed to different Kubernetes pods, essentially batching the consolidations as opposed to what it was before. This will reduce the time to complete consolidation tasks and reduce the risk of an API run getting stuck as the table building logic will now be in batches.
  
  Our CS team will be testing and implementing these improvements with customers, so please speak with your Customer Success Lead. (ref 15189)

One AI

• Group Attrition Regression Recipe - A new Recipe that predicts the amount of attrition for each of a defined set of groups has been added to One AI. The Group Attrition Regression Recipe is the first regression recipe featured in One AI. Regression problems involve predicting numerical outcomes rather than the categorical outcomes predicted in classification problems such as the other attrition risk recipes which predict whether or not each individual will terminate employment. The sections in this recipe differ from other recipes due the unique nature of performing a regression analysis on groups. Please note that only Generative Attributes can be used as features in this recipe. (ref 2581)
  
  
  
  
• A group selector was added to the Group Attrition Regression and the Custom / Advanced Recipes. This selector can be found in the section of the recipe where the population is defined and provides an interface for defining groups from dimension levels. When used in the Custom / Advanced Recipe along with a Regression problem type selection, any type of group regression analysis can be configured. (ref 3027)
• The error messaging in One AI machine learning runs has been refined to highlight when there are not enough features presented to the model to make reliable predictions. This message can be found on the Messages tab of errored runs where less than 5 features made it through cleaning. (ref 3066)

Minor Improvements & Bugs Squashed

• We are improving our load-balancing algorithm for queries that are running on Redshift to queue up queries more effectively, instead of sending them all to Redshift at once. (ref 19164)
• There were some instances where pages did not leverage the entire screen length but instead had a fixed height. We have made it possible now for pages to use the entire screen length. (ref 19480)
• We have resolved an issue where clicking on an invalid chip within the Define panel resulted in an Error being thrown. Now, the invalid chip is  not clickable, ensuring a smoother user experience and unnecessary errors. (ref 19328)
• Made a small improvement to the Admin > Users Page. Sorting now also works on the date columns, when deactivated users are OFF. (ref 19479)
• We have made improvements to allow widgets that get corrupted on a Storyboard tile to be deleted by a user. Previously, if an error occurred there was no menu option to delete the affected widget. (ref 19672)