I need to create a subset of data to analyze for my clients.
Specifically, I need to create a standard deviation list.
Weed out outliers (Z-Score over 5) and then recalculate on new list.
The list is extremely dynamic so please do not advise to weed out in load script as the combinations of dimensions are vast.
This is my calc for score:
ZScore=
(
Avg(Cost)
-
//National Avg
aggr( NODISTINCT avg ( {< VISN,Station,Facility,[Division Name] >}
(Cost)
) , Clinic )
)
/
//National STD DEV
aggr( NODISTINCT Stdev ( {< VISN,Station,Facility,[Division Name] >}
(Cost)
) , Clinic )
Figure my results are:
[Division Name] Clinic Zscore
1 x .5
2 y -1
3 y 10
...
My goal is to weed out the 10 and then recalculate Zscore.
Good luck! And thanks for your help.