Semi-Additive Measure Aggregation
hi,
i'm tearing hair (what's left of it!) out trying achieve in ssas 2005, , wonder if can before go mad! basic scenario i'm trying model is thus:
funds, made of number of:
investments, made of number of:
allocations, represent unit holdings.
now there may many allocations per investment, , each of these allocations (hence unit holdings) may change @ various points in time. fact table i'm using has entries representing current unit holding @ given date. example of fact table i'm using:
date fund id investmentid allocationid unitholding
2008-08-01 1 1 1 10
2008-08-03 1 1 1 20 (representing 10 unit holding increase)
2008-08-07 1 1 1 15 (representing 5 unit holding decrease)
2008-08-01 1 1 7 30
2008-08-15 1 1 7 10 (representing 20 unit holding decrease)
2008-08-21 1 2 9 15
2008-08-25 1 2 9 10 (representing a 5 unit holding decrease)
hence, if i wish find unit holdings given day, interested in lastnonempty value unitholding measure. instance, if want see unit holding allocation 1 on, say, 2008-08-10, value of 15. fine. however, want achieve, if want see unitholding @ lower level of granularity, investment, hope see sum of lastnonempty values each constituent allocation. in example, investmentid1 on 2008-08-20, hope see unitholding of 15 + 10 = 25. seeing, value of 10, since, guess, lastnonempty value investmentid 1. likewise upward levels of granularity.
so, in summary, i'm trying achieve measure uses lastnonempty aggregation @ highest level of granularity, these values summed @ lower granularities in hierarchy. have thought pretty standard functionality (say aggregating number of items held in stock across number of sites @ point in time, say), i'm feeling bit mr. thicky having particularly thick day @ moment , getting fast, amazingly welcome!
many thanks,
rob
if correct, trying have different aggregations measure @ different levels of dimension. can't think of solution except using calculated member.
SQL Server > SQL Server Analysis Services
Comments
Post a Comment