Hi all,
I’m working on building a horizontal bar chart in AWS QuickSight with the following goal:
- Each bar represents a specific check type (e.g., Annotation, Software Editor, Timestamp, etc.)
- The height of each bar reflects the total number of documents that have a ‘WARNING’ status for that check
- Each bar should be clickable, setting a parameter (e.g.,
pCheckDisplay) that is used to filter or control downstream visuals on the same sheet
Dataset structure:
The dataset is in wide format. Each row represents a document, and each check type is stored in its own column.
For example, columns like annotation_check_status, software_editor_check_status, and so on, contain values such as 'PASS' or 'WARNING'.
Approaches we have tried:
1. Calculated fields per check type + combo chart
We created calculated fields that return a 1 if a particular check type has a 'WARNING' and 0 otherwise. These were added as separate measures in a combo chart.
- Result: Accurate counts per check type
- Limitation: Combo charts do not support parameter actions on individual measures, so bar clicks cannot trigger interactions
2. Pivot table using calculated fields
We added each check type’s 'WARNING' count as a separate measure in a pivot table.
In this setup, the calculated fields are added to the Values field well, and we use a dummy dimension (e.g., a constant like “All”) to align them in rows.
- Result: Accurate counts are displayed per check type
- Limitation: Clicking on a pivot row does not trigger a parameter action, because pivot tables in QuickSight do not support interactivity at the measure level
3. Single calculated field using nested ifelse to assign labels
We attempted to create a single field that returns the name of the check type if its status is 'WARNING', and used this field as the Y-axis in a standard bar chart.
- Result: This approach enables click-based parameter setting
- Limitation: It only returns the first matched check type per row, so it undercounts when multiple check types have
'WARNING'in the same row
4. Dual dataset approach (wide + pivoted)
We created one dataset in its original wide format and another manually pivoted dataset to simulate a row-based structure for the bar chart.
These were both used in the same analysis and linked through a shared parameter (pCheckDisplay).
- Result: The pivoted dataset allowed us to build a clean, accurate bar chart with clickable bars
- Limitation: While clicking a bar set the parameter correctly, the rest of the sheet using the wide-format dataset did not respond to the parameter. Cross-dataset parameter interactions did not trigger the expected filtering behavior
Key limitations encountered:
- Combo and pivot charts do not support parameter actions at the measure level
- Using a single
ifelse-based label only captures one warning per row, which leads to undercounts - Coupling two datasets (one wide, one pivoted) does not result in consistent sheet-wide interactivity
- QuickSight does not support unpivoting wide-format data natively in the UI
What we’re looking for:
Is there a way in QuickSight to:
- Build a single horizontal bar chart from wide-format data (without SQL reshaping)
- Display accurate
'WARNING'counts for each check type - Enable bar clicks to set a parameter and trigger filtering or visibility across other visuals on the sheet