Hi,

I am analyzing the data of a survey. I have data of respondents and a
weight factor for each respondent. For the measure of unweighted
respondents I can use an aggrate function "distinct count". To
calculate the weighted number of respondents, I 'll need an aggrate
function "distinct sum" but Analysis server doesn't support the
"distinct sum" function.

RespId dim1 dim2 ... dim(n) weight measure1 ... measure(m)
234 . . . 0.62 . .
234 . . . 0.62 . .
768 . . . 0.45 . .
834 . . . 1.36 . .
834 . . . 1.36 . .

unweighted #respondents = 3 ( respId's {234, 768, 834} *no
duplicates*)

weighted #respondents = 0.62 + 0.45 + 1.36 = 2.43 ( sum of weight for
{234, 768, 834} )

Is there a solution (not depending on the other dimensions) for the
unweighted #respondents measure?

Beer

Hi,

my experience at analyzing data from some marketing&research surveys, where
you always have weight, is that you have to prepare your data before.
You can use distinct count, but in your case you need a group by statement.

So, I suggest that you prepare your data (fact table) in that way that you
measure are already prepared, as unweighted and weighted values.


Bye,
Andrej




"B.Molleman" <b.molleman (AT) marketresponse (DOT) nl> wrote