I have this MDX query:
with member [Measures].[count] as '[Measures].[NUMBER]'
** member [Measures].[total] as 'sum({[RECORD_TYPE.VALUES].[All VALUEs].children})' **
member [Measures].[% records] as '([Measures].[count] / [Measures].[total])', format_string = "0.0%"
select {[Measures].[count], [Measures].[% records], [Measures].[total]} ON COLUMNS,
[RECORD_TYPE.VALUE].[1] ON ROWS
from [RECORDS]
where Crossjoin({[CITY].[LONDON]}, {([DATE.DATE].[20171101] : [DATE.DATE].[20171130])})
I have found out that this [total] member only counts filtered records by date (records data between 20171101 and 20171130). If I switch this total to:
member [Measures].[total] as 'sum({[DATE.DATE].[All DATEs].children})'
I am getting all records in cube, regardless of city.
I would like this total to be the total number of records for [CITY].[LONDON] in cube, no matter the record type. Could you please tell me how do I do this?