brawer / osmdiffs / 42026755740

94%

Build Type

push

github

Committed by web-flow

Commit Message

feat(conflate): Emit OSM-only rows for deletion candidates (#876)

* feat(conflate): Emit OSM-only rows for deletion candidates

An OSM feature that no ATP feature matched now gets a row of its own in
conflated.parquet, with atp null, when a strong signal says ATP would
list it if it still existed:

- its brand/network/operator:wikidata names something ATP has features
  of the same kind for, and that we can match;
- it is a transit stop whose Swiss IDs (ch: IFOPT, 85 uic_ref) are all
  unknown to ATP;
- it is a transit stop in Switzerland without IDs, far from any stop in
  the Swiss register.

On Swiss data, this adds 2,358 rows. See
docs/notes/2026-11-05-deletion-candidates.md.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttfb2JREgH38uFM1gbohBJ

* feat(conflate): Count brand coverage per country

Look up countries with the country-boundaries crate instead of a
Natural Earth outline of Switzerland, which got Konstanz and the
enclaves wrong. A brand now counts as covered by ATP only in the
countries where ATP has features of it. The boundary data is
OSM-derived (ODbL) and is listed as an input in the provenance BOM.

On Swiss data, OSM-only rows go from 2,358 to 2,252.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttfb2JREgH38uFM1gbohBJ

---------

Co-authored-by: Claude <noreply@anthropic.com>

Coverage Stats

448 of 480 new or added lines in 8 files covered. (93.33%)

1 existing line in 1 file now uncovered.

13999 of 14869 relevant lines covered (94.15%)

2568.64 hits per line