stdev in MDX

[justinmarkham]

Hi all,
Has anyone here had any degree of success in using the stdev function in MS OLAP.
I have a field already in the fact table which I would like to calculate the stdev as a measure, like if it was an aggregate function. However the formula stdev (SET [, Numeric Expression]) does seem to be slightly irritating.
I've tried creating a dimension based on the field and then using the Dimension.members property for the set and have had little success.

The Data is along the lines of a survey response which can map to text responses. eg
Response Response mapping Question
1 Very Dissatisfied 1
2 Very Dissatisfied 1
3 Very Dissatisfied 1
4 Somewhat dissatisfied 1
5 Somewhat Satisfied 1
6 Somewhat Satisfied 1
7 Somewhat Satisfied 1
8 Very Satisfied 1
9 Very Satisfied 1
1 Yes 2
2 No 2

so if question 1 is selected I can see the stdev of all answers 1 through 9 or 1 through 3 if Very dissatisfied is also selected as a column or slicer.

Technet and MS help is kinda vague on this issue so I would appreciate any help and experience anyone has to offer.

Thanks in advance
Justin Markham.

 

[Guest_imported]

Hi!
I use LinRegSlope function which has very similary
syntax ( Set, NumericExpression, RankExpression ).
To calc linregslope you must have x-axis of data (which is
usually a timeline and y-axis - values of points. So:

LinRegSlope ( [Time Set], ( tuple of rest dimensions ), Rank( [Time].CurrentMember, [Time Set]))

Rank returns number of position current member in [Time set]. It starts from 1. This parameter is optional.
[Time Set] is set of choosed time dimension members.
(tuple) has one member from each dimension. Eg.

LinRegSlope ( [Time Set], ( [Client].[ClientX],
[Product].[Product Y],
[Measures].[Sales] ),
Rank([Time].Currentmember, [Time Set]) )

I hope it will help you!

Best regards
Grzegorz Stolecki
stolecki@consorg.com.pl