Skip to contents

Known hazards in the NCCS metadata name and address files

Source: R/md-sources.R
md_meta_caveats.Rd

Two things about meta_names / meta_addresses have to be settled before either is trusted as a variant source, and both are visible in the first rows of the files:

Details

  1. ORG_NAME_ALL is comma-concatenated and appears truncated. The first data row reads "SECOND BAPTIST CHURCH, PALMER SECOND BAPTIST CHURCH, PALMER SECOND BAPTIST CHU" - several names joined by commas, with the last one cut mid-word. Splitting on the comma is ambiguous because organisation names contain commas, and a truncated tail would enter the database as a name that never existed. Quantify the truncation rate before ingesting; a variant source that invents strings is worse than no variant source.

  2. ORG_ADDRESS_ALL is comma-joined as street,city,state,zip, which collides with commas inside street lines. Prefer address_resolved_crosswalk, which has already done this parse and records each address spell as its own row with a vintage range.

EIN2 is not a defect - it is the canonical identifier for this project. See md_ein(). An earlier revision of this note had that backwards.