Thank you for your contribution kaitejohnson 🚀! Your website is ready for download 👉 here 👈!
_{(The artifact expires on 2025-04-06T10:07:26Z. 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 (c50944e).
Report is 9 commits behind head on main.

@@            Coverage Diff            @@
##              main       #63   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           11        12    +1     
  Lines          305       334   +29     
=========================================
+ Hits           305       334   +29     

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

Approving but noting that there is now a naming collision between get_nowcast_df and get_prob_nowcast_df I think which do different things.

Is there? So the PR to generate_probabilistic_nowcast_df #55 actually creates two functions: convert_reporting_square_to_df() and nowcast_list_to_df() (which have the functionality of generating a probabilistic nowcast where a "nowcast" is the full reporting square)

I think the larger issue that I ran into when reorganising the methods is that I think we want to clearly define:

• reporting square with observed and imputed point estimates (I've been calling this a point nowcast)
• reporting square with observed and expected observed estimates (I've been calling this an expected observed nowcast)
• a vector of expected observed total counts at each reference time (this is being generated in this PR, so we are implicitly calling it a nowcast).

If we call the summarised by reference time thing a nowcast, we should clearly distinguish something of that form to the reporting square objects.

Would you say a "nowcast" is typically defined as the sum of the elements across delays, and the matrices are described in the more verbose way as say reporting triangle, reporting square, etc.?

Ah it didn't make it in.

nowcast_list_to_df()

I think this is still quite a confusing conflict as it sounds like the same thing.

reporting square with observed and imputed point estimates (I've been calling this a point nowcast)
reporting square with observed and expected observed estimates (I've been calling this an expected observed nowcast)

What is the difference between these two things? Without context they sound like the same thing?

If we call the summarised by reference time thing a nowcast, we should clearly distinguish something of that form to the reporting square objects.

This is the thing that in literature (i.e.) Johannes paper is what is being talked about in terms of a nowcast.

The additional complexity in the above is there are versions of all of these that do and don't include the observed data where it is available and there are use cases for both (most applied users what the data fused versions most of the time with the pure posterior prediction being mostly for model checking).

My usual call is to stick to the common bayesian naming scheme but here that won't work due to it not being Bayesian.

the summarised by reference time thing a nowcast, w

I am being a pedant here but I think we want to use aggregate or summed here or something as summarise is probably going to mislead people.

I'm not sure this PR is the place to have this discussion so maybe it calls for a meta issue w/ a table or something so we can get an overview and have a think?

What is the difference between these two things? Without context they sound like the same thing?

One is summarised by reference time (so just indexed by reference time) e.g. could be a vector, the other is indexed by reference time and report time (a matrix)

Yeah I will make a meta-issue with a table to distinguish between these things. However, for this PR I think its reasonable to still change the name if its confusing (which I believe it is currently). What about aggregate_by_reference_time() instead of get_nowcast_df()?

Made a new issue #74

Going to change the name to aggregate_by_ref_time()
Latest function flow: