Archive
Highlighted

How can I make a field that contains a time series window group of 3 events?

New Member

Is there a way I can group a window of 3 time points and add it as a field with the last two remaining being ignored?
I'm trying to classify time series patterns using a support vector machine with the Splunk MLTK and I'm unsure how to get the data in these windows.

(ex) My data has the _time and amount fields, and I would like to add the windows field:
1. _time amount windows
2. XX:12:XX 6 [ 6, 8, 4]
3. XX:13 XX 8 [8, 4, 4]
4. XX:14:XX 4 [4, 4, 3]
5. XX:15:XX 4 [ 4, 3, 2]
6. XX:16:XX 3 ...`
7. XX:17:XX 2 ...
`

Tags (1)
0 Karma
Highlighted

Re: How can I make a field that contains a time series window group of 3 events?

Ultra Champion
| makeresults count=2
| streamstats count
| eval _time=if(count=2,relative_time(_time,"-2d@d"),relative_time(_time,"-1d@d")) 
| makecontinuous span=1h _time
| eval amount=random() % 10 +1 
| table _time amount
`comment("from here, the logic")`
| reverse
| streamstats window=3 list(amount) as windows 
| reverse
| eval windows="[ ".mvjoin(windows," ")." ]"

Hi, @andrewhnguyen
how about this?

0 Karma