I refer to the outlier command
https://docs.splunk.com/Documentation/Splunk/7.0.4/SearchReference/Outlier
*Is there a way to draw the line of where the cutoff point for outliers is? *
I can play with the paramiters like this ... | outlier action=remove uselower=true param=1.6 |
, which moves the cutoff point for outliers, but I don't get to see the line exactly.
How can I finetune my outlier approach and or the param, because currently my outlier is cutting off datapoints that I do not want it to cut off.
I can also try to draw the line myself, but this is a very first attempt.
... | eventstats median(attempts) as med_att median(successfullAttempts) | eval x=med_att-(0.25*med_att)
which gives this pic
From the docs:
Filtering is based on the
inter-quartile range (IQR), which is
computed from the difference between
the 25th percentile and 75th
percentile values of the numeric
fields. If the value of a field in an
event is less than (25th percentile) -
param*IQR or greater than (75th
percentile) + param*IQR , that field
is transformed or that event is
removed based on the action parameter.