Histogram (bars) with counts as bins

[tsullivan (Tim Sullivan)]

@EnzoL

• You gave a search query that can be typed in in the Kibana console, and in fact returns "buckets" that count the number of participants for each meeting. Can this query be used for a Kibana visualization (as some background code)?

I looked around for quite a bit, and I couldn't find an aggregation type in Elasticsearch that can count the unique values of an aggregation output. I'll ask some more and see if I missed something in the documentation: Aggregations | Elasticsearch Guide [6.4] | Elastic

@EnzoL

• You talked about "pipeline aggregation" in Elasticsearch. Is this something that could be done in a Kibana visualization also (after all we want to draw bars, and not do console queries)?

Possibilities with pipeline aggregation also refers to the previous question. I'm not sure if there is a type of pipeline that can pipe the output of one aggregation and count the distinct values that came in.

@EnzoL

You said "If you don't have numParticipants as a field in your data, you always have the option of pre-processing that calculation at index time". Yes, we do not have that field. But I have no idea how this could be calculated "at index time" (do you mean in Logstash?). Would we need the Elasticsearch filter plug-in (since we do not have that available)?

I think you could use the Elasticsearch input plugin (not filter plugin) in Logstash: Elasticsearch input plugin | Logstash Reference [8.11] | Elastic. The input query would be the aggregation, and you could re-index the results into a new index, then use the new index in Kibana for the visualization.

If you don't have that plugin for Logstash and are not able to get it, you can achieve the same with some custom scripts in a language like Python or NodeJS to run a query, format the results into new documents, and index those into Elasticsearch.