PAL- outlier detection - PAL_ANOMALY_DETECTION function

Hi there,

I am using the PAL function PAL_ANOMALY_DETECTION for detecting outliers in retail sales data. I played around a bit with the parameters of this function. No matter what I ever try, I always get 1/10th amount of records in the result table of a given time series (e.g. 62 for 628, 31 for 314 etc.).

I would expect the parameter OUTLIER_PERCENTAGE to control this amount but all inputs I tried had no influence on this behavior.

Any suggestions or ideas?

Cheers

Alexander