2086010 - Result not making sense? - Information regarding Rounding, Averaging and Annualisation onsite - WFP

• Have a Result that is not making sense?

• Is the result correct at the monthly and quarterly level, but wrong at the annual level? (Annualisation)
• Measure annualisation is calculated as follows:
• ((Sum(all available months of data))/(Count(all available months of data))) * 12
• For example, annualising 7 months of Total Termination data gives the following result:
• ((Total Terminations (January to July) = 16,437)/7)*12 = 28,178

• Is rounding impacting the result?
  Example
  77/73 = 1.05479... (when calculated on a calculator), which as a percentage result would be 105.5% or as a whole number 106%
  Result showing onsite is - 105.2% Why?
  In this instance the denominator and numerator are being annualised BEFORE the calculation occurs, and then rounded for appearances onsite (showing as 77/73)
  The calculation however, uses a true figure from the cube, not the rounded figure showing onsite. (so may be (76.79/73.01) = 105.18559...
  Onsite would display as - 77/73=105.2%

Two different results when averaging same set of data?

I ran 2 queries per actual average unscheduled Absence per Employee - Sick Leave (FTE) over FY 08/09:

• Query 1: Unscheduled Absence per Employee - Sick Leave (FTE) - All Org Units, All Genders, All Ages, All Organisation Tenures, All Regions, Period: 08/09

• Query 2: Unscheduled Absence per Employee - Sick Leave (FTE) - All Org Units, All Organisation Tenures, All Regions, 08/09 - this is the same query as above but I have removed the Dimensions

All Genders and All Ages.

• The average of all data (ascertained by highlighting all figures in the excel table) in Query 1 is 8.44, whereas the average of all data in Query 2 is 6.77!
• This is best explained by simplifying the query to using just Tenure for a particular organisational unit.

• If you use excel to average the nodes under All Org Tenures
• (6.51 + 8.43 + 4.19 + 6.51 + 6.82)/5, the result is 6.49.

• This is a different result to the All Org Tenures result of 6.67.
• This difference in results occurs because weighting has not been taken into account. If all samples are of equal size, then an arithmetic average of the average will be correct. If sample size is different then it will be incorrect when compared with the weighted average.

• In this example, each of the All Org Tenures nodes has a different (FTE) sample size (11.21, 12.70, 11.45 etc). So to find the true average, you need to add the All Org Tenure nodes for Unscheduled Days Absent and then divide by the addition of the All Org Tenure nodes for FTE. E.g.:
  (73 + 107 + 48 + 160 + 514) / (11.21 + 12.7 + 11.45 + 24.57 + 75.4) = 6.67

KBA , LOD-SF-WFA-WA , Workforce Anaytics , How To