Solution:
Yes you can! Though it's good to note the behaviour of the tool when you are in this situation as there are times blanks in datasets are expected, and sometimes, blanks in dataset may not be relevant to a specific analysis.
Date columns
For date columns with blanks, If that date column is the default date, Auto Insights will not pick up the row with blank dates in it for any of its analysis.
Measure columns
For measure columns with blanks, Auto insights will treat that as 0 and aggregate the column accordingly. When calculating Averages, there is the option to 'Include blanks' if you wish to keep those specific rows in your analysis. This check box is displayed at the bottom left of the Measure picker in the Query Bar (found in Mission/Search) and is only available when Averages are selected. By default, this checkbox is unchecked.
Segment columns
For segment columns with blanks, Auto Insights will group the analysis as 'Unlabelled' as a default and will not be shown as part of the key call outs in 'What Caused This' even though it might be the largest change. Instead Auto Insights will call out the next largest contributor in that segment that contributed to the change as the key driver.
This prevents datasets with columns that are expected to have blanks in them to have misguided insights. E.g. Service Ticket dataset with 'Assigned Owner' column having blanks is expected if it's a newly created ticket that has not been triaged yet and should not be called out as a main contributor for the largest increase, instead the next largest is of significance.
When you add a segment with blanks into the 'breakdown' dropdown in the Query bar (found in Missions/Search), Auto Insights will generate an Unknown Values story and calculate the significance of the blanks in your analysis. In this screenshot it has highlighted that the breakdown analysis identified has 14.7% blanks, giving the user the ability to make a judgement call if the generated insights are suitable.
Solved! Go to Solution.
