Skip to contents

Surface confusable type pairs through the taxonomy and the data

Source: R/11-fx-similarity.R
fx_sibling_report.Rd

Compares every pair of types through two lenses – taxonomic distance (fx_path_distance()) and signature distance (fx_signature_distance()) – and flags where they disagree. hidden_risk pairs (far in the taxonomy but near in signature space) are the dangerous set that needs hard negatives banked against each other; safe_split pairs are siblings the data tells apart easily.

Usage

fx_sibling_report(
  sigs,
  paths,
  labels = names(paths),
  weights = SIGNATURE_WEIGHTS,
  path_near = 0.34,
  sig_near = 0.1
)

Arguments

sigs

A named list (or matrix of rows) of signatures.

paths

Ontology paths, aligned to sigs.

labels

Row labels (default names(paths)).

weights

Signature weights (default SIGNATURE_WEIGHTS).

path_near, sig_near

Thresholds for taxonomic / signature nearness.

Value

A data frame of pairs with distances and a status column (both, hidden_risk, safe_split, unrelated).

See also

fx_path_distance(), fx_signature_distance()