Read the alias table, one row per (EIN, name)

Collapses the (ein, name, city, state, zip5) grain to (ein, name), keeping the best-corroborated address for each pairing, and drops names that are ambiguous across EINs — a name used for two different organizations cannot identify one of them.

Usage

pf_schedi_alias(
  db = file.path(pf_schedi_dir(), "schedi_grants.duckdb"),
  min_filers = 2L,
  drop_ambiguous = TRUE
)

Arguments

db

Path to the DuckDB file.

min_filers

Minimum distinct grantmakers. 1 admits everything (86% precision), 2 is the sensible floor (95%), 5 is auto-accept grade (98%).

drop_ambiguous

Drop names that map to more than one EIN.

Value

A data.table, one row per (ein, name_norm).