I am not aware of anything called "outlier group" in APO. 
Outlier correction is an optional function you can use 
during statistical forecasting to clean your historical 
data of points that are much higher/lower than the rest of 
your numbers. You can influence the "sensitivity" of the 
function by changing the so-called sigma factor (unlike the 
name suggests it is NOT standard deviation) which defines 
the threshold over/under which numbers are 
considered "outliers".