Distinct count to exclude NULL

问题

Is there a way to create a calculated measure that can do a distinct count, but excluding the null values? For example, if I have: 10, 20, 20, null; the distinct count is 3, but can I have the answer as 2?

回答1:

put a where condition to where the measure is not null

回答2:

you could try this:

with cte ( "counter" ) as
( 
    select 
        1 as "counter" 
    from 
        "YOURTABLE" 
    group by 
        "COLUMNNAME" 
    having ( "COLUMNNAME" IS NOT NULL )
)
select SUM( "counter" ) from cte

@ edit:

quite easier variant:

SELECT COUNT( DISTINCT "COLUMNNAME" ) FROM "TABLENAME"

回答3:

have the same issue - it is amazing microsoft ignores it...looks like the only solution is to build a view for distinct count measure group

http://www.sqlservercentral.com/Forums/Topic810069-17-1.aspx#bm1213461

回答4:

Thank you for all the answers. However, here I found a solution that works best for my need: http://richardlees.blogspot.com/2008/10/alternative-to-physical-distinct-count.html

By creating a Calculated Member using the following:

count(nonempty({[DimName].[HierarchyName].[LevelName].members-[DimName].[HierarchyName][All].UNKNOWNMEMBER},[Measures].[MyMeasure]))

回答5:

simply add where value is not null and value "!=''" that should do the trick

来源：https://stackoverflow.com/questions/13914039/distinct-count-to-exclude-null