Retool PM here. This sounds like a good use case for Retool! (I built something similar to join performance data with metadata using Retool in my last role).
While we don't currently support using Python or R in apps, we do have something new shipping soon that'll make it easier to run heavier statistical lifting server side. If you're interested in a beta access, or even pairing on a build session, let me know at jane@retool.com.
Also, I think pre-calculating values and storing them in a database table is a smart idea if your velocity is increasing and you want folks to be able to access reports themselves. Your users will get a snappier app and you'll have a historical record of experiment results to reference.
To answer my own question: My last big XSLT project was for a video curation app startup back in 2010-ish. When the iPad came out and every company needed an app, you could proxy app calls via Charles to access unprotected (and unstandardized) XML feeds of video content. For the better part of a month I downloaded every app with video, looked for accessible API feeds and wrote XSLT transformations and XPath to standardize for consumption by our app. At times it was rote, but I learned you can gleam a lot about a company when you inspect their internal APIs.
While we don't currently support using Python or R in apps, we do have something new shipping soon that'll make it easier to run heavier statistical lifting server side. If you're interested in a beta access, or even pairing on a build session, let me know at jane@retool.com.
Also, I think pre-calculating values and storing them in a database table is a smart idea if your velocity is increasing and you want folks to be able to access reports themselves. Your users will get a snappier app and you'll have a historical record of experiment results to reference.