Hi everyone, I would like to create home assistant charts. Occasionally (for unknown reasons) there is some reading error from a modbus sensor. I would like not to consider values that are too different from the previous ones. Eg. If a value is more than 20% different from the average of the last 5 it should not be considered. I would like to eliminate the spikes that are seen in the attached graph. There’s a way?
To remove spikes you can use Graphite functions for data filtering like removeAboveValue/Percentile and removeBelowValue/Percentile (Functions — Graphite 1.2.0 documentation). It won’t do exactly what you’re looking for but may help to filter out unwanted values.
Hi, thank you!
where i can find the graphite functions? consider that i’m using grafana on influxdb on Home assistant (HASSIO) installation
Best regards and thanks so much
Oh I thought you were using Graphite (the post is in Graphite category). If you’re using InfluxDB I think you can simply pass AND "field" < threshold to the WHERE clause to filter out spikes:
without filtering
with filtering:
(Only display values over a certain threshold - #4 by katy - Dashboards - InfluxData Community Forums)
Hi,
I revive this old post. The suggestion of @piotrjamroz does not consider (for example) the average of the queried data.
It would be nice to be able to hide data that get too far from the average, without hard-coding thresholds that may suppress valuable data.
Does anyone have an idea?
Thanks!
1 Like