Add “Number of Unique Values” as a Summary Statistic in Graph Builder Color Role

[BHarris]

What inspired this wish list request? 

In many of my workflows, I use heat-maps to analyze data where individual entities (such as parts, units, or serial numbers) can appear multiple times due to repeated tests or observations. For example, a single part may fail across several test runs, resulting in multiple rows tied to the same part ID.

When using Graph Builder, I often want to visualize how many unique parts are associated with each category (e.g., failure mode, test condition, location). However, when placing a variable in the Color role, the available summary statistics (such as count or sum) operate on rows rather than distinct entities. This makes it difficult to quickly answer questions like “How many distinct parts failed in each category?” without pre-aggregating the data or creating additional columns/tables.

Currently, the workaround involves creating a summary table or using scripting to compute unique counts beforehand, which interrupts exploratory analysis and adds unnecessary friction.

What is the improvement you would like to see? 

I would like to see an additional summary statistic option in Graph Builder, specifically for variables placed in roles like Color (and potentially Size, etc.):

“Number of Unique Values” (or “N Unique”)

This option would compute the count of distinct values for a selected column within each grouping defined by the graph.

Why is this idea important? 

This capability would significantly improve exploratory data analysis by allowing users to directly visualize distinct entity counts without needing to preprocess the data.

Key benefits:

• Reduces friction – eliminates the need to create intermediate summary tables or scripts
• Improves accuracy – avoids accidental misinterpretation when row counts differ from entity counts
• Enhances interactivity – supports real-time exploration in Graph Builder without leaving the tool
• Widely applicable – useful across many domains (reliability, manufacturing, testing, survey data, etc.)

In my specific use case, this would make it much easier to distinguish between “number of failures” and “number of failed parts,” which are often very different and equally important metrics.

Overall, this feels like a natural and high-impact addition that would make Graph Builder even more powerful for real-world data analysis.

[hogi]

You can easily get this via a Transform column.

Instead of putting “parameter” into the color rule — and suffering from a non‑existing N Unique in the statistics —
create a Transform Column of parameter: Col N Unique(:parameter) and use this transform column for coloring.

This workaround was already possible before, but with JMP 19 EVERYTHING changed: Scoping!
Just activate scoping at the level you want, and Graph Builder will calculate the statistics down to exactly that level — and, even more importantly:

it will automatically adjust the statistics when the user changes the settings in Graph Builder.

Besides the new Data Grid, this is the biggest improvement in JMP 19 — so great! :rocket: