Proper Scoring Rules
Calculates the logarithmic, quadratic/Brier and spherical score from a model with binary or count outcome.
performance_score(model, verbose = TRUE)
model |
Model with binary or count outcome. |
verbose |
Toggle off warnings. |
Proper scoring rules can be used to evaluate the quality of model
predictions and model fit. performance_score() calculates the logarithmic,
quadratic/Brier and spherical scoring rules. The spherical rule takes values
in the interval [0, 1], with values closer to 1 indicating a more
accurate model, and the logarithmic rule in the interval [-Inf, 0],
with values closer to 0 indicating a more accurate model.
For stan_lmer() and stan_glmer() models, the predicted values
are based on posterior_predict(), instead of predict(). Thus,
results may differ more than expected from their non-Bayesian counterparts
in lme4.
A list with three elements, the logarithmic, quadratic/Brier and spherical score.
Code is partially based on GLMMadaptive::scoring_rules().
Carvalho, A. (2016). An overview of applications of proper scoring rules. Decision Analysis 13, 223–242. doi: 10.1287/deca.2016.0337
## Dobson (1990) Page 93: Randomized Controlled Trial :
counts <- c(18, 17, 15, 20, 10, 20, 25, 13, 12)
outcome <- gl(3, 1, 9)
treatment <- gl(3, 3)
model <- glm(counts ~ outcome + treatment, family = poisson())
performance_score(model)
## Not run:
if (require("glmmTMB")) {
data(Salamanders)
model <- glmmTMB(
count ~ spp + mined + (1 | site),
zi = ~ spp + mined,
family = nbinom2(),
data = Salamanders
)
performance_score(model)
}
## End(Not run)Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.