Matrix stats on string fields

Hi,

I read up and played a bit with the new Matrix Stats aggregation that was committed to master recently. It is a nice addition for a number of use-cases and works quite well for numeric values.

What I am wondering now is if there is a way to apply the same concept on string fields. As an example if you have log-entries that contain the type of browser and the operating system where they originated, there will usually be a high correlation of mobile browsers to mobile operating systems.

Is there a way I can run queries that will allow me to find such statistical correlations?

I can probably use some machine learning tools like Mahout, but it would be nice being able to use only Elasticsearch for this.

Thanks... Dominik.

Unfortunately Matrix stats is limited to numeric fields at the moment. Adding support for String fields is something we're still considering.