Aggregation & Analysis

Question

I'm attempting to use Discoverer to create a rolling 12-month attrition report. It works fine, up to the point of trying to create a average headcount for each month in the current 12-month period over the last 12 months' each. The problem I'm encountering involves the use of the MIN() function in selecting active employees in each month, mostly due to a data-cleansing issue, which I'd hoped to bypass. Because some invididuals have two "data conversion" records - i.e., they were converted to the new database and an additional, subsequent record re-used what should have been a unique action reason - I need to test their MIN(position start date) so as to then use their actual start against the first record, whereas I can go on to use their position start to capture their FTE for any subsequent active records.
So I created a calculation, Min Start, to hold the earliest start date for each employee:
MIN(Position.Start.Date) OVER (PARTITION BY Employee.Number ORDER BY Position.Start.Date RANGE BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING)
Then I test the status of each position for each month in the report and return the FTE where it tests as active through month-end.
CASE WHEN Position.Action.Reason = 'CNV' THEN
CASE WHEN Position.Start.Date = Min Start THEN
CASE WHEN Hire.Date <= Month.End THEN
CASE WHEN Termination.Date is NULL THEN FTE
WHEN Termination.Date >= Month.End THEN FTE END END
WHEN Position.Start.Date <= Month.END
CASE WHEN Hire.Date <= Month.End THEN
CASE WHEN Termination.Date is NULL THEN FTE
WHEN Termination.Date >= Month.End THEN FTE END END END
WHEN Position.Start.Date <= Month.End THEN
CASE WHEN Termination.Date is NULL THEN FTE
WHEN Termination.Date >= Month.End THEN FTE END END
However, I can't SUM this because nesting isn't permitted and I can't average the sum because Aggregation of Analytic functions is not allowed either. I need a different approach. Data is always going to be dirty, so coding to account for such problems means I can perform my reporting requirements without interruption for clean-ups. (I know that keeping the data clean is best and highlighting such problems brings the attention of managers and staff to rectifying and avoiding such problems, but I still need to get the results out.)
I guess I need a new approach, but I'm at the end of every google search and textbook I can locate. Any suggestions would be very welcome.
Edited by: 814208 on 21/11/2010 19:14

Answer 1

Updated: For the analysis authorization, for these characteristics, in addition to the specific values, there is an entry for ":" and this did not resolve the issue.
Authorization Check  
  Detail Check for InfoProvider ZBL_13IT  
  Preprocessing:  
Selection Checked for Consistency, Preprocessed and Supplemented As Needed
Subselection (Technical SUBNR) 1
Check Node Definitions and Value Authorizations...
Node- and Value Authorizations Are OK
End of Preprocessing
  Main Check:  
  Subselection (Technical SUBNR) 1  
Supplementation of Selection for Aggregated Characteristics
  Check Added for Aggregation Authorization:     0COMP_CODE  
  Check Added for Aggregation Authorization:     0PLANT  
Following Set Is Checked  Comparison with Following Authorized Set  Result  Restmenge 
Characteristic  Contents 
0PLANT
0SALESORG
0TCAACTVT
0COMP_CODE
SQL Format:
COMP_CODE = ':'
AND PLANT = ':'
AND SALES
ORG = '0403'
AND TCAACTVT = '03'
Characteristic  Contents 
0PLANT  I EQ IC01
I EQ IC02
I EQ IC03
I EQ IE02
I EQ IE04
I EQ IZ01
I EQ MC01
I EQ ME01
I EQ :
0SALESORG  I EQ 0301
I EQ 0341
I EQ :
0TCAACTVT  I EQ 03
I EQ :
0COMP_CODE  I EQ 0327
I EQ 0341
I EQ :
Not Authorized   
All Authorizations Tested
  Message EYE007: You Do Not Have Sufficient Authorization  
  No Sufficient Authorization for This Subselection (SUBNR)  
Following CHANMIDs Are Affected:
142 ( 0COMP_CODE )
189 ( 0PLANT )
192 ( 0SALESORG )
  Authorization Check Complete