WinsorizedMean¶

Mean after winsorizing extreme values — the lowest and highest fraction of the data are replaced by the nearest retained value before averaging.

$ wo 'WinsorizedMean[{1, 2, 3, 4, 5, 6, 7, 8, 9, 100}, 0.2]'
11/2

A pair {flow, fhigh} winsorizes the two ends by different fractions.

$ wo 'WinsorizedMean[Range[10], {2/10, 1/10}]'
57/10

The fraction defaults to 0.05, so a one-argument call takes nothing off a list of ten:

$ wo 'WinsorizedMean[{1, 2, 3, 4, 5, 6, 7, 8, 9, 1000}]'
209/2

An association is averaged over its values:

$ wo 'WinsorizedMean[<|"x" -> 1, "y" -> 2, "z" -> 1000|>, 1/3]'
2