Summarise One Design Member from the Shared hc Shards

The per-scenario fan-in used by ssd_design_targets(): reads the (shared) hc shards under dir_hc, filters to the member's serving hc task identities (hc_ids) - its own cells, plus the coincident ci = TRUE cell serving each of its ci = FALSE cells under ssd_design_targets()'s ci routing - projects out the dists/samples list-columns at the DuckDB level, and writes the member's compact summary at path. When the design aggregates differing readouts into a shared cell, proportion/est_method narrow the maximal computed set back to the member's requested readout rows. The read, filter, projection, and write all happen inside DuckDB (never collecting into R), mirroring ssd_summarise().

Usage

ssd_summarise_member(
  dir_hc,
  hc_ids,
  path,
  proportion = NULL,
  est_method = NULL
)

Arguments

dir_hc

The (shared) hc results root of the member's seed group.

hc_ids

The member's serving hc task identities (hc_ids) to keep.

path

The output Parquet path for the member's compact summary (the format = "file" contract).

proportion

Optional proportion values to keep (the member's requested proportions when the design aggregates a wider union into the shared cell); NULL (the default) keeps every proportion.

est_method

Optional est_method values to keep (the member's requested methods when the design aggregates a wider union); NULL (the default) keeps every method.

Value

path.

See also

ssd_design_targets(), ssd_summarise_design(), ssd_summarise().