scoringutils
scoringutils copied to clipboard
Utilities for Scoring and Assessing Predictions
Create function `as_point()` that converts sample- or quantile-based forecasts into point forecasts
Meta-issue: #433. Next issue: #434 We would like to be able to compute point forecasts for quantile- and sample-based forecasts. The cleanest way to do this would be to have...
The function `range_long_to_quantile()` was previously used in `score_quantile()`. It transforms forecasts from a long interval format to a quantile format. Depending on whether we want to support the interval format...
And maybe also think about whether that function should be exported at all... I think the main reason it's exported is because it's an S3 method.
In his review of #762, @seabbs made a few good comments and suggestions, like - removing the author - adding a few references - more
The current version of the manuscript should be a package Vignette so we can easily link to it in the docs.
Just we've done with `validate_forecast()` in #778 we should add a `verbose` argument for `as_forecast()`
Code refactoring: Deduplicate code in `get_coverage()` and `interval_coverage_deviation_quantile()`
Meta issue: Code refactoring: #459 Functions share a lot of code that could be turned into helper functions
As described in #559, some functions cause an object to lose its `forecast_*` class by calling `as.data.table()`. We should go through all functions to make sure this isn't happening /...
This meta issue tracks ideas to refactor the code to make it more efficient / cleaner. **Outstanding issues:** - [ ] #454 - [ ] #411 - [ ] #445...
- [ ] check function documentation is correct and clean - [x] check all relevant functions have input checks - [ ] decide on pkgdown categories (#481) - [ ]...