Visual error in Tree map

I have tree map visualization created that is throwing the following error when using a specific filter (client filter):


To exemplify: I am using a group by we can call “group by 1” that always contains data. The measures for the color and the size of the visualizations are “measure 1” and “measure 2”.
To avoid duplicity we use the same analysis for each client, but there are some clients that do not have any values for “measure 1” and “measure 2”. So when I filter for a client in specific that does not have values for “measure 1” and “measure 2”, we are getting blanks in the treemap as shown in image.

Now that does not happen with a Filled map, where we have the same situation where we do have the grouping values available but no measure. The Filled map shows the group by (in this case “state”) values and just colors every state with the same color intensity:

Filtering by the same client with the same conditions two different visuals are reacting differently, and tree map visual is showing all blank. Is there a fast way to solve this maintaining the structure of the analysis as is right now?

Hello @joana , I think I understand the issue you are facing. Maybe rather than trying to get something to show if no data is present you could instead hide the visual when those clients are selected. It wouldn’t display any helpful data for the user and you are already using the free form layout, so conditional rule logic can be added to that specific visual to hide it when that scenario is present.

If that doesn’t fulfill your requirements, could you send a little more detail regarding the desired outcome?

1 Like

Hi Dylan! Thank you for your response
The way the dataset is constructed we should still be able to see the values for the grouping being used and then have everything colored and sized the same size and color. For example in the map all states and countries have the same color intensity and we can still see the geographical data point there.
I am aware of the conditional hiding functionality but unfortunately this happens to most of the visuals in the dashboard, so if there is no other alternative I’d have to analyze other options.

Thanks again!

Hello @joana ! I am understanding the scenario a bit more.

I am not certain this would work, but the issue could be the visual determining the expected value is null and displaying the error message in response to that. I think there could be a solution that would require you to create a calculated field that would check if the value you are using for Size and Color is not null to use that value or return 0. That way there is a value returning to the visual no matter what and it should at least maintain the display.

Something along the lines of ifelse(isNotNull({measure 1}), {measure 1}, ‘0’}) and replacing that calculated field with the value of measure 1 and measure 2 in those field wells.

1 Like

Hello @joana , did my response direct you to a solution for your QuickSight question?

Hey @DylanM I tried it but it doesn’t work, I guess that the queries that qs is doing in the backend to fill the visual aren’t returning any data for the metrics presented (even when using a calculated field to set them to 0) so in those specific cases the visual doesn’t even fill with it’s attribute values.
Thanks again for your answers, I’ll keep you up to date when I find an alternative!

@joana I appreciate the update! I will dig into this more as well and see if I can find another possible solution.

@joana , I think the big difference between the 2 visuals you were looking at is the group by value for the filled map determines a visual representation so even if your 2nd field well is empty it will continue to show the map.

As for the tree map, if you filter out data for either size or color on the tree map, the visual loses a required field to view the visual and won’t be able to display.

Did my response help you resolve your QuickSight question?

Hey Dylan, thanks for the clarifications here! I’ve decided that best move here is to change the approach/visuals based on your feedback and research.

I’m glad you figured out a way forward! Good luck with your dashboards!