TSVB terms visualization

[oshtman (Andreas)]

Hi,
I'm fairly new to Kibana and I'm attempting to setup a visualization where I have fields of the form
("my_field_name" + "custom_part"):
my_field_name_custom_part1=true
my_field_name_custom_part1=false
... etc
What I want to achieve is to be able to group the field by terms and apply a filter only showing the count of desired variants of the field (e.g. I only want to visualize all occurrences of my_field_custom_part1).

However, grouping by terms in e.g. a TSVB table and later applying a filter in the dashboard only appears to allow sorting by either "count" or "asc/desc" order of the field which terms was applied to. Is there any way to limit the table results to only show my_field_custom_part1?
Note: The fields I want to filter on are selected from a custom drop down list

I'm not sure if its clear what I want to achieve but shoot me any questions if not

[flash1293 (Joe Reuter)]

I'm not 100% sure about the visualization you want to achieve, but using the "Aggregation based" visualization types it's possible to order the terms by a custom metric and you can use a metric based on any field in there. Does that help?

I think screenshots of the current state of your dashboard and a description of what you want to change would help me understanding your case.

In general, it's not possible to dynamically aggregate multiple fields based on a common name prefix in Kibana - you always have to pick the field to operate on. There are some ways around this, for example you could ingest a single document per dynamic part - instead of

{
  my_field_name_custom_part1: true,
  my_field_name_custom_part2: true
}

you could do

{ key: "my_field_name_custom_part1", value: true }
{ key: "my_field_name_custom_part2", value: true }

[oshtman (Andreas)]

Hi, and thanks for your initial reply.
Re-structuring the data as you suggested did help with setting up most of the desired visualizations.
Regarding the ordering however, I do use filter ratios in TSVB. As it appears, filter ratios are not available in visualizations created from the "Aggregation" list. Is there any other way to ensure I only display what the global filter is currently set to? Seems like a pretty standard thing and I'm sure its possible somehow, maybe I'm just coming at the problem from the wrong direction.
Some images that might help:

Dropdown selecting desired field experiments.key.name as filter:

Table created with TSVB using filter ratios, which ignores my gobal filter of expriments.key.name. Should show: "new_design" entry only

[flash1293 (Joe Reuter)]

I see, if you need filter ratios TSVB is the right choice at the moment.

TSVB visualizations should respect the global filter on dashboards. Is it possible there is a single document containing multiple values for for experiments.key? Keep in mind that Elasticsearch is flattening out nested objects, so a document like this

experiments: [
{ key: "new_promotion_pitch", value: 4 },
{ key: "new_design", value: 7 },
]

is interpreted like this in aggregations:

experiments.key: ["new_promotion_pitch", "new_design"],
experiments.value: [4, 7]

So it's not tying the key to the value in nested objects. To visualize that, the individual experiments should be separate documents.

[oshtman (Andreas)]

That was actually the miss on my side, and the cause of the ordering not working as expected.
Thanks again for your help!