My MDX script:
SELECT NON EMPTY { [Measures].[Fact Emp Violation Count] } ON COLUMNS, NON EMPTY { ([Dim Date].[Calendar].[Year].ALLMEMBERS * [Dim Warning Level].[Violance Hierarchies].[Violance Type].ALLMEMBERS ) } ON ROWS FROM ( SELECT ( { [Dim Date].[Calendar].[Year].&[2015] } ) ON COLUMNS FROM [Violation])
My tables:
I want to group based on violence type first letters
My desired outputs is:
Year | Violence Type | Fact Emp Violation Count
CY 2015 | 1 | 19 + 18 + 2 + 23 + (violence type begins with '1')
CY 2015 | 2 | 11 + 4 + 1 + (violence type begins with '2')
Thanks.
First you need to do a small cube design change for the solution to be optimal. In the DSV, add a named column and call it [New Violance Type]
. Logic for that field is straightforward -
=left([Violance Type], 1)
My computing it before hand and storing that in the cube does wonders to your MDX query response time. Although it can be computed as a part of MDX too, but it decreases the performance.
Add this field to the same dimension as a new attribute hierarchy - [dim warning level].[New Violance Type]
Then you can use the below MDX -
select non empty { [measures].[fact emp violation count] } on columns,
non empty { ([dim date].[calendar].[year].allmembers * [dim warning level].[New Violance Type].allmembers ) } on rows
from ( select ( { [dim date].[calendar].[year].&[2015] } ) on columns from [violation])