Navigation menu

A big part of the app experience is the navigation between sheets. With our new navigation menu, you can style it as you prefer. Set to a top menu, a side panel, or as a pop up drawer. Pick any colors and choose the font that you desire.

Navigation menu

Tab container

The new tab container is an improvement the old container object with a new name and styling settings for labels, tabs and icons. Key features include:

• Label font styling: Family, size, color, alignment.

• Label icon size, color and alignment.

• Tab background, selected, and hover color settings. Color by fixed colors or by expression.

• Styling for labels, tabs and icons, both for the whole container and per tab.

• Support for label width by pixel or percent.

• Support for longer labels.

• Settings for vertical or horizontal tabs.

• Improved menu experience.

Tab container

Visualization improvements

This week includes the following enhancements:

Straight table

Chart exploration in the new straight table is now supported in layout container and the new tab container. The new straight table also gets word wrap of headers and content.

Straight table

Pivot table

The new pivot table gets range selection.

Layout container

The layout container is now supported in custom tooltips.

Holding shift while resizing the layout container will keep charts in same place and size.

Scatter plot

Scatter plot gets options for bubble outline color and support for reference lines with slope.

Creating scatter plots

Improvements to intelligent model optimization

The processing and training workflow for intelligent model optimization has been enhanced with new capabilities. Intelligent model optimization is activated by default in new ML experiments, and offers automated refinement of machine learning models created with AutoML.

Continuous numeric features will be binned into buckets to re-balance the distribution of data. This new feature will be tested to see if the quality of the model is impacted with and without the transformation.

For numeric features that are identified as skewed, the values will have a power transform applied to generate new features for model training.

Intelligent model optimization