Thomas Karbe
05/29/2024, 9:04 PMFact Conversions
with a Conversion Created Day
and a Dim Features
with a Feature Release Date
, where a conversion can be related to multiple features via the map table. Since the Feature dimension is very small, this is no big deal computation-wise. Now I want to build a Metric Conversion Count for Features not older than 6M
that just looks for each conversion if there is at least one feature related to it that was released max 6M before the conversion creation. I later want to show this metric either as an absolute or in relation to total conversions. I tried to do it via the BY keyword, but so far none of my metric attempts were applicable. They never show up in the available selection…Michal Hauzírek
05/30/2024, 5:53 AMFact conversions
and Map conversion to feature
to a many-to-many edge in the GoodData logical model. This part of the model seems to be an ideal candidate for that, with no complications.
Then any filter applied to features
(including a date filter) will get propagated to to Fact conversions
via the map
table.Thomas Karbe
05/30/2024, 7:40 AMThomas Karbe
05/30/2024, 12:21 PMConversion Count for Features not older than 6M
or in fact any other metric combining the two data sets e.g. Min days between conversion creation and release of a feature
. Could you please provide some examples that help me build those?Michal Hauzírek
06/02/2024, 11:06 AMSELECT COUNT(Conversion Key) WHERE Month/Year (Feature release date) >= THIS-6
Michal Hauzírek
06/02/2024, 11:15 AMSELECT MIN(
SELECT
(SELECT Date (Feature Release Date) - Date (Conversion Created Date) BY Map Key)
BY Feature Key, Conversion Key, ALL OTHER
)
the very inner metric performs the difference of the dates on the Map Key level. The next level with BY Feature Key, Conversion Key, ALL OTHER
seems to be needed there to make it work properly in case both of the keys are not present in the report. And the topmost MIN is just to get minimum.
I hope this helps.Thomas Karbe
06/03/2024, 5:10 PM