Shape changes the shape of each point to represent that point’s value or category for the assigned feature.
Shape accepts all feature types but is best used with categorical features that have six or fewer unique values. Categorical features with more than six values will have the values with the lowest number of members combined into an “Other” category. Numerical features will be divided into six bins, although Shape does not have as many bin customization options as the Color dimension.
Using Different Sets of Shapes
Shapes: This dropdown selection, included in the toolbar, allows users to choose between 3 options for the display of data points:
- Default 3D – default for the best combination of quality and performance
- Legacy 3D – previous 3D rendering method for legacy support
- These have a slightly higher performance cost than the Default 3D shapes
- Point Cloud – shapeless objects for each data point
- When using Point Cloud, the user has the ability to map features to Size or Shape but these dimensions are not visible on the plot
- Point Cloud has the lowest performance cost, and can be used to reduce machine load when visualizing very large datasets (millions of data points)
- It can help show Color trends more clearly when there are many data points very close to each other on the plot