Analyse a single outcome feature with all estimators (internal)

Thin wrapper around .estimate_total_feature() that extracts vectors from the dat list format used by the simulation pipeline.

Usage

.analyze_feature(dat, f, W_avg, W_valid = NULL)

Arguments

dat

Dataset list (from generate_toy_data / run_single_iteration).

f

Feature (column) index into dat$Y, dat$W, dat$G.

W_avg

Row means of the full negative-control panel (for COCA).

W_valid

Optional: validity-screened W matrix (n x q) for the matrix-bridge PGC. If NULL, uses the full W panel.

Value

Data frame of five rows (one per method) or NULL if too few cases.