ref_flag

Usage

ref_flag(scores, threshold = 2)

These are threshold exceedances, not abnormalities. With many well-calibrated independent outcomes, about 0.0455 * p rows are expected to exceed |z| > 2 by chance.

Arguments

Argument	Description
scores	A ref_scores table.
threshold	Absolute Z threshold.

Value

The score table with exceedance and fdr columns, of class ref_flags. The attributes expected_exceedances and observed_exceedances give the chance expectation under calibration and the observed count.

Examples

sc <- as_scores(distributional::dist_normal(0, 1), c(0.2, 2.6, -3.1))
sc$.in_sample <- FALSE
ref_flag(sc)

<ref_flags> 2 exceedances (expected 0.137)

# A tibble: 3 × 11
  observed median  centile      z tail_prob tail_surprisal residual log_density
     <dbl>  <dbl>    <dbl>  <dbl>     <dbl>          <dbl>    <dbl>       <dbl>
1      0.2      0 0.579     0.200   0.841            0.173      0.2      -0.939
2      2.6      0 0.995     2.6     0.00932          4.68       2.6      -4.30
3     -3.1      0 0.000968 -3.1     0.00194          6.25      -3.1      -5.72
# ℹ 3 more variables: .in_sample <lgl>, exceedance <lgl>, fdr <dbl>