Total variation distance similarity score per categorical column

Usage

tvd_similarity(real, synthetic, meta)

Arguments

real: A data frame of real data.
synthetic: A data frame of synthetic data.
meta: An rsdv_metadata object.

Value

A tibble with columns column (chr) and score (dbl, 0–1, higher = better).

Examples

# \donttest{
syn   <- gaussian_copula_synthesizer(metadata(adult_income)) |> fit(adult_income)
synth <- sample(syn, n = 500)
tvd_similarity(adult_income, synth, metadata(adult_income))
#> # A tibble: 9 × 2
#>   column         score
#>   <chr>          <dbl>
#> 1 workclass      0.975
#> 2 education      0.958
#> 3 marital_status 0.98 
#> 4 occupation     0.936
#> 5 relationship   0.948
#> 6 race           0.99 
#> 7 sex            0.998
#> 8 native_country 0.971
#> 9 income         0.99 
# }