Usage of Array in a Kibana Data Table : issue with Count behaving as a Unique Count

Elastic Stack Kibana
1

Hi,

A am facing an issue when trying to visualise data coming from an array.
For instance, my array 'Tableau_test' contains ths below

PUT /discuss_test3/_doc/1
{
"Tableau_test":[
"Moto",
"Voiture",
"Voiture",
"Velo",
"Velo",
"Velo",
"Marche",
"Marche",
"Marche",
"Marche"]
}

I am expecting to see the below in a Kibana Data Table :slight_smile:
Marche 4
Velo 3
Voiture 2
Moto 1

But the 'count' metrics seems to behave as a 'Unique count' metrics...

cf. below

image
image1219×349 20.6 KB

If someone can explain/help
Thanks !
David

2

Try Sum instead count

3

Thanks. Already tried ; failing with "The index pattern discuss_test3 does not contain any of the following compatible field types: number" error

4

What you showed in your example is a single document with an array field, which gets to the core of how Lucene (which is underneath Elasticsearch) represents data.

This visualizations is showing the right thing- it shows that there is a single document that contains each value.

5

The "cardinality" aggregation is based on unique documents, so to get the behavior you were expecting, I would recommend indexing more than one document.

6

Thanks Wylie, it helps a bit.
I tried to change my data source ;
my data source is now:
PUT /discuss_test4/_doc/1
{
"Tableau_test":[
{ "mode":"Moto", "nombre" : 1},
{ "mode":"Voiture", "nombre" : 2},
{ "mode":"Vélo", "nombre" : 3},
{ "mode":"Marche", "nombre" : 4}
]
}

But I am still failing to get a visualisation in Kibana showing the below:
Marche 4
Velo 3
Voiture 2
Moto 1

Cheers

7

Yes, nested objects are even less supported in Elastic and Kibana. Using multiple documents is your best bet.

8

Yeah. Thanks. We will do so.
Cheers