I referenced Splunk documentation on finding outliers below.
Why is there a need for moving a window? My use case is that I collect stats say port loading factor over the last 5 mins and calculate upper threshold ie. mean+3*sigma. If any stats in this 5 mins goes above the upper threshold, I declare that it is an outlier. Then I repeat everything for the stats that comes in the next 5mins. So there is no moving window. Am I doing it right? Or I need the moving window. Thanks for your advice.
| inputlookup quote.csv
| head 500
| eval _time=(round(strptime(time, "%Y-%m-%d %H:%M:%SZ")))
| streamstats window=100 avg("price") as
avg stdev("price") as stdev
| eval lowerBound=(avg-stdev*2)
| eval upperBound=(avg+stdev*2)
| eval isOutlier=if('price' < lowerBound
OR 'price' > upperBound, 1, 0)
I think that if you are using streamstats window=100 to compute avg and stdev, your moving window is exactly 100 events. This should correspond with your 5 minutes window.
In case you want to use all data as reference for finding outliers from avg and stdev, you should use eventstats to compute avg and stdev of all data in your search.