Hi everyone I am trying to filter some outliers but, the methods used just filter the upper outliers. The data is very simple and I know that my outlier selection is subjetive.
I am also attaching my attempts
Upper and Lower Bound Outlier.knwf (39.3 KB)
Hi, in your example the first Quantil (0.25) is part of the “lower Outliers”. Even you set the k = 0 in the numeric outlier detection, the numbers 5 and 6 are not defined as outliers.
So, what you define as lower outliers are no outliers according to a “statistic theory”.
In your case you should use a k-medoids cluster nodeforum_outlier.knwf (16.6 KB)
with k=3 (number of clusters) and define cluster 1 as lower and cluster 3 as upper outliers (sorting the number ASC).
Thank you Tommy excelent approach
This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.