Thank you for this example. This is very much along the lines of what I'm looking for. Previously I used ODK aggregate and connected directly to the underlying postgres database via metabase to achieve similar functionality. However, I've now moved to ODK central on docker, where the postgres database is hidden in docker and requires some hacking to expose for outside connections. Metabase does not support Odata feeds, so it is nice to see that Redash does. However, I see that you go back to using a postgresql datasource for your more advanced example. Is it still ultimately easier use a database datasource? For instance, my ODK forms always have repeats in them, which then require different types of sql joins for analysis. Can I do that using Odata and Redash or am I best off spending my time figuring out how to expose the ODK central postgres database for outside connections so I can just do sql queries directly?