Issue #846 - Implement scoring for ordinal forecasts

[nikosbosse]

Description

This PR closes #846.

• implements a new class, forecast_ordinal
• implements the log score and the ranked probability score for ordinal forecasts

The PR should probably have a few more tests at least for the RPS. @nickreich @elray1 do you happen to have a validated version/example somewhere that we can use?

I also haven't implemented the PIT yet, but I think that's a separate part of the codebase anyway (get_pit_histogram).

There is a small change in the way a vignette is run conditional on ggidst being present that you can probably just ignore. Not sure why it worked before... See more info here: #917.

Checklist

• My PR is based on a package issue and I have explicitly linked it.
• I have included the target issue or issues in the PR title as follows: issue-number: PR title
• I have tested my changes locally.
• I have added or updated unit tests where necessary.
• I have updated the documentation if required.
• I have built the package locally and run rebuilt docs using roxygen2.
• My code follows the established coding standards and I have run lintr::lint_package() to check for style issues introduced by my changes.
• I have added a news item linked to this PR.
• I have reviewed CI checks for this PR and addressed them as far as I am able.

[elray1]

Thanks for putting this together! Quick response to your question about existing implementations of RPS: I don't know of any formal, well-tested implementations. The best I've put together is a quick proof of concept.

[nikosbosse]

Ah just realised that scoringRules has the RPS - maybe we should use that https://github.com/FK83/scoringRules/blob/master/R/rps.R

[nikosbosse]

@elray1 updated the code and edited unit tests for the rps. It now uses the scoringRules version. I think/hope I got the logic for reordering the columns correct, but would appreciate a second pair of eyes

[seabbs]

This looks good to me. I'm not totally across the rps so it might be a good idea for someone else to check

[elray1]

I read through the code and tried out a few worked examples. The functionality all looks good/correct. I had a few comments on documentation and example setup, as well as some suggestions for opportunities for refactoring/sharing code between the nominal and ordinal forecast classes which I'm perfectly happy if you ignore.

[nikosbosse]

Most recent updates:

• refactored assert_input_nominal and assert_input_ordinal to use a common assert_input_categorical
• replaced logs_ordinal and logs_nominal with a common logs_categorical and added a deprecation warning.
• accepted suggestions from @elray1 related to docs

[elray1]

lgtm