Interesting approach. Any thoughts on how this self blend would impact performance in a large data set?
A shame that yet again, Tableau requires some non-ideal workaround.
1 of 1 people found this helpful
Blend performance would typically depend on the cardinality
(the number of unique value combinations) of the dimensions to blend on.
Figures at about tens of thousands would noticeably impact the performance.
Besides, filters on the blended fields from Secondary (such as the [Error Type] from the above)
are restricting the Marks on a view, so filtering could potentially help with the performance.
Hope this could help understanding.