Tidy a kernR Test Result Into a One-Row-Per-Term Data Frame

Turns a kernR verdict object into a flat, broom-style summary data.frame: one row per tested term, with stable, documented columns. The method is registered against the tidy() generic (re-exported by broom), so broom::tidy(result) and generics::tidy(result) both dispatch here.

Usage

# S3 method for class 'kernel_test_result'
tidy(x, ...)

Arguments

x: A kernel_test_result (the class returned by hsic_test(), mmd_test(), bd_hsic_test(), and friends) or a subclass such as mmd_ppc.
...: Currently unused; present for generic compatibility.

Value

A one-row data.frame with columns:

term: Character. The quantity tested – the test method (for example "bd-HSIC", "MMD").
statistic: Numeric. The observed test statistic.
p.value: Numeric. The permutation p-value. Note the broom-canonical dot, distinct from the result's native p_value field.
n: Integer. The sample size the statistic was computed on.
n_permutations: Integer. Number of permutations in the null.
ess: Numeric. Effective sample size of the importance weights, or NA for unweighted tests.

Subclasses contribute extra columns where they carry extra fields: an mmd_ppc result adds surprise_bits and reject.

Details

This is the supported accessor for downstream code. Reaching into the result list by field name is fragile: kernR's native field is p_value (with an underscore), whereas the broom convention – followed here – is p.value (with a dot). Use tidy() and read p.value rather than guessing at the raw field name.

Examples

set.seed(1)
x <- matrix(rnorm(200L), 100L, 2L)
y <- x[, 1L] + rnorm(100L)
res <- hsic_test(x, y, n_permutations = 199L, seed = 1L)
tidy(res)
#>   term  statistic p.value   n n_permutations ess
#> 1 HSIC 0.01021759   0.005 100            199  NA