Influxdb: Should support moving averages aggregate function
Should be able to calculate moving averages of different types. Simple, weighted, exponential, etc. Need to come up with specific syntax and stuff, but for now there's more reading here http://en.wikipedia.org/wiki/Moving_average
pauldix
All 89 comments
:+1: EMA functions would be great!
nichdiekuh
on 31 Jan 2014
Comments from #396 (which is a duplicate of this issue)
It would be great to have Influx calculate a (simple) moving average (or median, or maybe even percentiles).
For example: SELECT movingAverage(value, $window) FROM my_series, where $window could be a number of points/events, or a time interval, like 1s or 2m.
jvshahid
on 3 Apr 2014
+1
spatil
on 19 May 2014
+1
huhongbo
on 22 Jul 2014
+1
clkao
on 17 Aug 2014
+1
sahilthapar
on 18 Aug 2014
+1
rmihael
on 19 Aug 2014
+1
selbyk
on 19 Sep 2014
+1
ybogdanov
on 16 Oct 2014
+1
shaharke
on 24 Nov 2014
+1, really need this. Should we think about it together?
smartynov
on 28 Nov 2014
+1 but as a downsampled continuous queries
select mavg(val) from temp_series group by time(1h) into temp_series.mavg.1h
mchmarny
on 18 Dec 2014
+1
DarrenMowat
on 18 Dec 2014
+1
ajcrowe
on 15 Jan 2015
+1
probepark
on 25 Jan 2015
+1
pbrit
on 4 Mar 2015
+1
tbillet
on 2 Apr 2015
+1
naneau
on 2 Apr 2015
+1
ebereketoglu
on 2 Apr 2015
I'd really like to have a go at this (and Holt winters forecasting). CQ seems the right way of doing these. They'd need access previous buckets/periods. Any tips would be appreciated. I'll start having a dig around
tcolgate
on 12 Apr 2015
+1
cemo
on 15 Apr 2015
window functions give the simple moving average for free and they are a really nice feature (http://www.postgresql.org/docs/9.4/static/tutorial-window.html).
haven't seen a database that provides EMA out of the box. usually it's a user defined function.
mbalkanloo
on 20 Apr 2015
:+1:
mooremo
on 23 Apr 2015
+1
digitalwm
on 30 Apr 2015
+1, moving average is something we would also appreciate.
Our use case is having metrics of raw monitoring data, let's say every minute pushing:
memory.<hostname>
free: <free_bytes>
total: <total_memory_bytes>
I was trying to have continuous query for pctFree_memory.
select (free * 100.0 / total) from /^memory\./ into pctFree_:series_name
This one on it's own does not work (the historical values are calculated but new ones are propagated with free and total; the calculated value is not there).
On the other note I was trying something like:
select mean(free * 100.0 / total) as pctFreeAvg5m from /^memory\./ where (time > now() - 5m) AND (time < now()) into pctFreeAvg5m_:series_name
But this one also does not work.
Can you perhaps point me into a right direction if such metrics are possible directly with continuous queries at all?
geronime
on 30 Apr 2015
+1 on this. One way mentioned on this page is
select mavg(val) from temp_series group by time(1h) into temp_series.mavg.1h
but that only gives a downsampled average, not a moving average. Does anyone have leads on this?
kespindler
on 16 May 2015
+1
jsfrerot
on 21 May 2015
+1
josh-padnick
on 22 May 2015
+1
nzjess
on 22 May 2015
+1
robochat
on 10 Jun 2015
+1
mmedeiros
on 11 Jun 2015
+1
MarekRychlik
on 25 Jun 2015
+1
ezeql
on 27 Jun 2015
+1
p4fg
on 27 Jun 2015
+1
andremann
on 1 Jul 2015
++
dmichael
on 6 Jul 2015
+1
zachkendrick
on 6 Jul 2015
+1
mipapo
on 17 Aug 2015
++
daijithegeek
on 31 Aug 2015
+1
bbinet
on 4 Sep 2015
+1
cagdascirit
on 22 Sep 2015
+1
markuspatrick
on 23 Sep 2015
+1 :-)
manoochehri
on 5 Oct 2015
+1
This is a useful function in Graphite, I need it in influxdb also.
lunxian
on 18 Oct 2015
+1
wzph
on 22 Oct 2015
bump
hillar
on 2 Nov 2015
+1
aadrian
on 2 Nov 2015
+1
smaddula
on 4 Nov 2015
+1
mecton
on 19 Nov 2015
+1
maklaut
on 20 Nov 2015
+1
rackunits
on 20 Nov 2015
+1
filippovitale
on 27 Nov 2015
+1
mariusoe
on 29 Nov 2015
+1
gilgamezh
on 11 Dec 2015
+1
burthen
on 22 Dec 2015
+1
jonseymour
on 24 Dec 2015
+2
ddaka
on 29 Dec 2015
+1
jcmartins
on 29 Dec 2015
+1
men1n2
on 29 Dec 2015
+1
leomaros
on 30 Dec 2015
+1
jlmcgehee21
on 7 Jan 2016
+1
aglagla
on 7 Jan 2016
+1
zas
on 29 Jan 2016
+1
cwegener
on 30 Jan 2016
+1
brumfb
on 5 Feb 2016
+1
mhiller
on 5 Feb 2016
+1
mrwulf
on 12 Feb 2016
:+1:
ojarva
on 15 Feb 2016
+1
j3ffrw
on 18 Feb 2016
+1
zapient
on 20 Feb 2016
+1
adrianlzt
on 23 Feb 2016
+1
pieroxy
on 28 Feb 2016
+1
jmaitrehenry
on 1 Mar 2016
unsubscribe
On Mar 1, 2016 11:24 AM, "Julien Maitrehenry" [email protected]
wrote:
+1
—
Reply to this email directly or view it on GitHub
https://github.com/influxdata/influxdb/issues/77#issuecomment-190820693.
mbalkanloo
on 2 Mar 2016
+1
thegreatjerboa
on 2 Mar 2016
unsubscribe
On Wed, Mar 2, 2016 at 3:49 PM, Jeff Schumacher [email protected]
wrote:
+1
—
Reply to this email directly or view it on GitHub
https://github.com/influxdata/influxdb/issues/77#issuecomment-191423281.
mbalkanloo
on 2 Mar 2016
@mbalkanloo that's not how you unsubscribe to an issue. You're adding more noise whenever you send that email. Follow the link to GH and unsubscribe to the issue using their interface.
pauldix
on 2 Mar 2016
+1
hailpam
on 8 Mar 2016
+1
tomazbracic
on 8 Mar 2016
As mentioned in my post to the mailing list we are experimenting with simplifying our open GitHub Issues. This feature request has been rolled into an aggregate issue for all function requests, so that we can close this issue until we are ready to work on it.
You may continue to make comments here. Closing the issue does not mean we are rejecting this idea.
beckettsean
on 18 Mar 2016
I'll be making a new issue for each component of this issue. The simple moving average aggregate is in #6112 and we hope to be able to get that into 0.12.0, but no promises.
I'm reclosing this issue for now.
jsternberg
on 24 Mar 2016
+1
k3gk3gk3g
on 9 Jul 2017
+1
szymic1
on 9 Aug 2017
+1
blanet
on 22 Sep 2017
+1
shawafts
on 5 Dec 2017
+1
dterg
on 30 Dec 2017
+1
alrghods
on 1 Mar 2018
+1
vincentbarat
on 23 May 2018
We have implemented some moving average functions. If you need a moving average function that isn't included in the latest version, please open a new issue so we can triage the issue.
Thanks. I'm going to lock this conversation to prevent additional +1 comments.
jsternberg
on 23 May 2018
Related issues
beckettsean
·
105Comments
beckettsean
·
44Comments
mvadu
·
60Comments
beckettsean
·
43Comments
TechniclabErdmann
·
80Comments