Ability to Filter/Remove Records After Review/Flagged

Hi, Community! I’m looking for ideas on developing an interactive dashboard for data quality reporting. The business use case would be to develop data quality reports that display records that don’t meet predefined quality standards (i.e. a field is null or a value is x std.d away from the mean). These records would then be displayed in QuickSight for further review; in some instances though the record might be deemed as valid and should no longer display on the report. An example would be a numeric value that is more than x std.d away from the mean but is determined to be valid. The challenge is how to create a report that enables a quality analyst to mark the record as valid and negate the record from reappearing on the report. Even more hopeful is that the review method would capture attributes about each record marked as valid.

There’s the potential to add fields at the source to flag said records but I’m hoping to develop something more dynamic and flexible downstream that mitigates having to engage data architects to update a data model.

Two ideas come to mind.

  1. Use conditional formatting to flag/highlight specific cases that meet requirements. This is easy to implement and quite flexible so any analyst could complete this task.
  1. Create a calculated field using a series of ifelse() statements to set the criteria and then add that field to the table to indicate what fields need to be reviewed.

Sounds like the first option might better suit your use case. It’s quite flexible and you can add multiple conditions.



so this could be a little bit complicated.

darcoli descriped a process to highlight rows in a table. We can use this to populate a parameter with values using the action feature.
In a calculated field you can test with ifelse, if the value of your primary key is in the parameter. If yes you get a 1, else a 0 and then you filter this field on 0. So we get only values, that were not selected as valid.

Now your analyst has to create a bookmark. This way the valid values keep being filtered out the next time the analyst opens the report.

If the analyst selects another value as valid, the bookmark needs to be overwritten.

Was this understandable? :smiley:


@thomask, haha, it was (kinda of). I feel like they will be very much a trial-and-error effort.