Specific SQL keyword

Hello Jean-Christophe,

The first case could be easily solved by “NOT IN” clause. Something like:

SELECT Learner WHERE USER_ACTION_FACT_TYPE  in ('Bookmark', 'Share', 'Comment', 'Like','ArticleView'  ) AND NOT in ('Training Session' )

Or you could easily filter out the values from the particular attribute later on when creating insight or report. 

Regarding the second case, unfortunately I cannot see any way how to achieve this via MAQL. That might be something that should be done on your end before loading the data into our platform. 

Feel free to check the bellow article for more information about SQL vs. MAQL:

https://help.gooddata.com/doc/enterprise/en/dashboards-and-insights/maql-analytical-query-language/maql-versus-sql/

Hello Jean-Christophe,

Thanks for your reply. 

Just for the clarification - should the MINUS mean in this case that you are for example trying to return only particular unique Learners which are present in (Bookmark, Share, Comment, Like,ArticleView) but excluding the same ones which are present in (Training Session)? To get rid of “duplicates” ones?

Or is the question really related only to COUNT of the total users/learners as you have suggested above? If thats the case, the bellow should work:

In MAQL, you can subtract metrics (please see Arithmetic Operations). This can solve the first part of your question. For example you can create two metrics:

1st_Metric:

SELECT COUNT(Learner) WHERE USER_ACTION_FACT_TYPE  in (Bookmark, Share, Comment, Like,ArticleView ) 

2nd_Metric:

SELECT COUNT (Learner) WHERE USER_ACTION_FACT_TYPE  in (Training Session)

And then subtract them within another metric:

SELECT  1st_Metric - 2nd_Metric

Regarding the other part of the question - Again, this can be definitely achieved by separated metrics already used above:

• Total of users that are only in (‘Training Session'):
  SELECT COUNT (Learner) WHERE USER_ACTION_FACT_TYPE  in (Training Session)
   
• Total of users that are only in ('Bookmark', 'Share', 'Comment', 'Like','ArticleView' ):
  SELECT COUNT(Learner) WHERE USER_ACTION_FACT_TYPE  in (Bookmark, Share, Comment, Like,ArticleView )
   
• Total of users that are in both:
  SELECT COUNT(Learner) WHERE USER_ACTION_FACT_TYPE  in (Bookmark, Share, Comment, Like,ArticleView, Training Session)

However, if this is not what you are looking for exactly, could you please elaborate a bit more in regards how is your LDM structured and also provide us with some particular example use-case and example of the result you would like to achieve? 

Thanks!