DMX Where Clause: Filtering According To The Adjusted Probability

I have a DMX query like this:




Code Snippet



select * from (
select flattened(*) from (
select att1, topcount(predict([Trans Predictor Unified], INCLUDE_STATISTICS), $Adjustedprobability, 7) as predictedstuff
from [Trans Predictor Model]
prediction join
SHAPE {openquery(DMSCS, 'select distinct CAST(att2 as nvarchar(100)) att1 from DMSCS.dbo.CartProducts order by att1 ')}
append
({openquery(DMSCS, 'select CAST(att2 as nvarchar(100)) att1 , att4, att5 as att3
from DMSCS.dbo.CartProducts order by att1 ')
}
relate [att1] to [att1]) as [Trans Predictor Unified]
as SHAPEQ
on [Trans Predictor Model].[Trans Predictor Unified].att3 = SHAPEQ.[Trans Predictor Unified].att3
) as s
) as t where [predictedstuff.$AdjustedProbability] > 0.5





It's working well. I would like to modify one thing. I would like to chang ethe constant in the where condition, so that it is configurable. That is, I would like to store the constant somewhere (SSAS or relational SQL). I was reading the DMX reference, but it doesn't provide much details about the where's "condition expression". And I looked at a document called "OLE DB for Data Mining Specification version 1.0" of July 2000, which does have in Appendix B the SELECT grammar. There it has

<expression> -> <value>
[...]
| ( SELECT <expression_list> FROM <expression> <where_clause>
[...]

<where_clause> -> WHERE <expression>

If I change the end to

where [predictedstuff.$AdjustedProbability] > (select 0.5 from [Trans Predictor Model] )

, however, just to force some form of query there I get a message saying "The specified column was not found in the context".

I'm running SQL Server 2005.

thanks,
Gustavo