How can I create graphs like the two in the picture, without using count/average/min/max etc.? I just want to plot the actual values of a field (e.g. kWh) on the y-axis and the time on the x-axis, and show this in my dashboard. Also, is it possible to show the anomaly detection from 'machine learning' on the dashboard?
There's not really any easy way right now to plot individual points.
Depending on the input rate of your data, you could zoom in the time range to the point where you only have 1 value in each interval. Then regardless or min, max, avg, you would see the only value at that time.
The general reason Kibana only shows aggregated results is because many cases of showing every data point would overwhelm a browser with the amount of data.
Experimental Vega visualization can do it. There's a sample that shows individual points like a scatter plot. But I'm not sure if you have a version of Kibana that includes that sample data or the Vega visualization type?
I think Canvas can also do it, but I don't see that your Kibana version has that. What version are you on?
Hi! Thanks for your reply. My version does have Vega Visualization, the screenshot is not my own. However, I'd like to have the anomaly detection that happens in Machine Learning on my dashboard. I've also set a watcher in 'management' to make a log when a condition is met, but is there a way to put these alerting logs on my dashboard as well?
You can't add ML anomaly charts to dashboards yet. Hopefully someday.
If you have watches creating logs, and those logs are indexed into your elasticsearch cluster, then you can;
- create an index pattern in Kibana for it
- open the index in Discover
- add specific fields of that log index
- create a saved search from that
- add the saved search to your dashboard
Hi, thanks! I'm trying to recreate the second link you sent but I don't really get it.
I uploaded some data in the Machine Learning part, then I created a job which multi metric viewer to detect anomalies. It shows me a graph with the anomalies detected, and also the colored map with squares. Then, I go management -> index pattern -> create index pattern -> and I select my created job. Then, I go to visualisation -> heat map, but I only get the fields my data contains. No 'influencers'. Btw, my data contains the value of voltage over time.
What does your Multi-metric job configuration look like? What does an example raw document of the voltage data over time look like?
Example of data:
timestamp text value unit
28-1-2020 08:00:08:940 energy 140 kWh
There are more columns in the actual data but I don't need them. Also, there are multiple different 'text' fields and also multiple units to measure (e.g. hours, L, V, A), but I am just using one unit right now to test if that works (so I deleted the other entries not being kWh).
Metric job configuration:
bucket span: 10m
That's it. It displays the value over time and detects anomalies.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.