Thank you for your contribution kaitejohnson 🚀! Your website is ready for download 👉 here 👈!
_{(The artifact expires on 2025-04-12T14:20:56Z. You can re-generate it by re-running the workflow here.)}

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 100.00%. Comparing base (3957643) to head (581a4d1).
Report is 11 commits behind head on main.

@@            Coverage Diff            @@
##              main       #69   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           11        11           
  Lines          305       325   +20     
=========================================
+ Hits           305       325   +20     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

I had a go at shortening the overview as I felt like it was a bit wordy. This might be a bit drastic.

The baselinenowcast model, initially developed as a reference for the COVID-19 hospital admissions nowcasting challenge in Germany (2021-2022), utilises reporting triangles of preliminary case counts and their delays. It applies a multiplicative approach, using empirically observed historical delay distributions to estimate yet-to-be-observed cases.
Users can specify whether delay distributions are estimated from the current reporting triangle, fully observed past data, or data from different strata. The model produces point nowcasts by "filling in" the reporting triangle.
Probabilistic nowcasts are generated using a negative binomial model with means from the point nowcast and dispersion parameters estimated from past nowcast errors. These parameters can also be flexibly sourced from current data, historical data, or alternative settings.

This looks good to me but the language here depends on our answers to #74 (because here you are using point nowcast to mean fill in the reporting triangle, which is also how I have used it, but in other places it means the estimate of the sum across delays for each reference time).

I think I would prefer to merge this now and then open a new PR to update all the docs and function names after we decide on the naming convention

sounds good