Confusion structure: what shape does "wrong" take?

Confusion structure: what shape does "wrong" take?

Usage

fx_benchmark_confusion(
  res,
  level = c("data_type", "semantic_family", "semantic_type")
)

Arguments

res

Output of fx_benchmark().

level

Ontology level to tabulate (default "data_type").

Value

A list with matrix (gold x predicted counts at level), wrong_shape (for each gold value, the distribution of the wrong predicted values, as proportions), and outcome (the per-gold-data_type hit / graceful / misroute mix, so graceful truncation is separated from a committed wrong branch).

See also

fx_benchmark()