Some Analytical DM Parameters

Summary

Description of some Analytical DM Parameters

Content

 

OutlierSensitivity is a parameter that defines how far (srd error units) the specific observation is from its near neighbors.
the lower the number the more aggressively the engine will detect potential outliers.

RemoveExtremeOutlier - If this is set to No - the engine will flag all the outliers as "suspicious as outliers " and if the forecasting method fails validation, it will then remove the suspicious points and will try to forecast with the smoothed data series.
If this parameter is set to yes, all suspicious outliers based on the criteria defined by OutlierStdErr (that should be >= OutlierSensitivity) are smoothed before the forecasting methods even try to do the initial forecast.