fs_identification_structure: Identification structure for one set of replicates

View source: R/fs_identification_summary.R

fs_identification_structureR Documentation

Identification structure for one set of replicates

Description

Classifies every detected replicate's realized rule against a planted anchor / partner / proxy triple, and returns covariate involvement, the structure composition, and classification accuracy.

Usage

fs_identification_structure(
  results,
  anchor,
  partner,
  proxy,
  confounders = NULL,
  sg_def_col = "sg_def",
  covs_col = "covs"
)

Arguments

results

A per-replicate results data frame, a bundle carrying results, or a path to an .rds holding either.

anchor

Character. The covariate the search is expected to recover.

partner

Character. The true second covariate of the planted region.

proxy

Character. A correlated covariate that may substitute for partner.

confounders

Character vector. The analysis covariate pool, for the involvement vector. Default NULL, which uses the three named roles.

sg_def_col, covs_col

Character. Columns holding the realized rule and its projected covariate names.

Value

A list with det_rate, n_detected, n_total, involvement, structure (a named share vector), and the four accuracy means.

The classification

Each detected replicate falls in exactly one of five mutually exclusive categories, tested in this order: the anchor is absent (⁠anchor missed⁠); the rule names the anchor and nothing else (⁠anchor only⁠); the rule pairs the anchor with the true partner; the rule pairs it with the proxy; or the rule pairs it with something else. The order matters – a rule naming both the partner and the proxy counts as the true pairing.


forestsearch documentation built on Nov. 15, 2026, 9:06 a.m.