Best Of

Ideally, your data would look like this:

Category Type Value

Budget Planned 343

Budget Carry-over 3432

…

The easiest way to get it in that format is with MagicETL. There's probably a more elegant way to do it with a dynamic unpivot, but to me, the easiest to understand is to just select the two columns, add a category column and stack the results on top of each other:

Once it's in that format, the stacked bar is easy:

If you transform your dataset

Then create a stacked bar using category on the xaxis, value on the yaxis, and type in series, you get this:

To transform your dataset, use Magic ETL.
Setup selected columns.
Add an Unpivot Tile. Drag the Unpivot Columns tile onto the canvas.
Unpivot Columns: Select budget, forecast, and actuals.
New Column for Key Names: Enter Category.
New Column for Values: Enter Value.
This will turn the three columns into two: Category and Value.

The label options on boxplots make them essentially useless, IMO. You can't even position them in the correct location (i.e. median goes on the median).

You could concatenate the N to the axis labels. Something like this:

Disadvantage is changing the names of the series makes cross-filtering between cards dice-ier.

I would love some better options to graph variation in Domo (or even just better boxplots).

If green is a subset of blue, I think the overlay bar is the appropriate graph. The advantage of the overlay is that even without figuring out the percent that you want, the raw values will be correct, so hovering and tips will work out of the box.

The default calculation for percent_of_category isn't necessarily wrong, it just isn't what you want for your situation. There isn't a way to tell Domo that your series is hierarchical, so it's assuming you want to add up the different groups in your series to comprise the category.

When aggregating the values for green, you can't really directly access the values for blue, since they are held on different rows than the values for green. What you can do though, is used a fixed function to find the total for the category (blue+green) and then subtract out the green values. That's what my beastmode is doing above:

sum(`Value`) / — find the numerator (total of greens)

(

sum(`Value`) fixed(by `Category` ) — this is the same denominator as percent_of_category (blue+green)

-sum(`Value`) —subtract out the greens

) —a denominator of blues

First, I don't think that's an overlay bar, because there's no option label totals in an overlay bar.

But if you were using an overlay bar, MichelleH is right that the best approach would be a BeastMode and a tooltip. It's an unusual percentage to get, but I think if you're using an overlay, I agree, it's what you want. Here's a BeastMode that gives the total for Blue and then Green/Blue for not Blue. Since we're mixing whole numbers and percents, I did the formatting in the BM, too:

Graphed as an overlay: