Proper Scoring Rules for Categorical Data

Calculates the logarithmic, quadratic/Brier and spherical based on a fitted mixed model for categorical data.

scoring_rules(object, newdata, newdata2 = NULL, max_count = 2000, 
    return_newdata = FALSE)

Arguments

object

an object inheriting from class "MixMod".

newdata

a data.frame based on which to estimate the random effect and calculate predictions. It should contain the response variable.

newdata2

a data.frame based on which to estimate the random effect and calculate predictions. It should contain the response variable.

max_count

numeric scalar denoting the maximum count up to which to calculate probabilities; this is relevant for count response data.

return_newdata

logical; if TRUE the values of the scoring rules are ruturned as extra columns of the newdata or newdata2 data.frame.

Value

A data.frame with (extra) columns the values of the logarithmic, quadratic and spherical scoring rules calculated based on the fitted model and the observed responses in newdata or newdata2.

References

Carvalho, A. (2016). An overview of applications of proper scoring rules. Decision Analysis 13, 223–242. doi:10.1287/deca.2016.0337

Author

Dimitris Rizopoulos d.rizopoulos@erasmusmc.nl

Examples

# \donttest{
NA
#> [1] NA
NA
#> [1] NA
NA
#> [1] NA
# }