search

FK83 / scoringRules

scoring rules to evaluate probabilistic forecasts
53 stars 16 forks source link

implement and export general DSS formula #37

Closed bastistician closed 5 years ago

bastistician commented 6 years ago

All the DSS functions eventually need to compute something like

(y - meanP)^2/varP + log(varP)

I think it would be cleaner to have a separate dss function which does just that, use it in the distribution-specific functions, and also export this basic function to enable the package user to compute the DSS for distributions not covered by the package.

FK83 commented 6 years ago

Thanks - we'll think about that.

The ensemble spread score (ESS) of Christensen et al (2015, http://onlinelibrary.wiley.com/doi/10.1002/qj.2375/full) which depends on the first three moments of the forecast distribution would probably be a worthwhile addition as well.

FK83 commented 5 years ago

Sorry - I forgot to close this one. @slerch implemented the functions dss_moments and ess_moments way back in 2017, see 330a3244bcb83636312d6512b52d2c1368f6a96f. Thanks again for the hint!