Outliers are values that lie outside an expected range. usually outliers occur because of some unexplicable or unstatistical reasons. Hence outliers would only disrupt a good forecast calculation based on standard values
so if you have vaues in a time series that go:- 10,12,13,17, 45,13,12,14,1
then its possible that 45 is an erratic value that can be called as an outlier. its also possible 0 is an outlier. the calculation part is to determine the allowed or expercted tolerance range.
once its been pinned as outliers and that their existence is not the normal behaviour of the time series, outliers need to be eliminated because their presence would change the forecast calculations
outliers are a standard statistical term and APO DP provides means to correct them. there is a recent thread here on this topic
Add comment