Stops matching
Matching Stages
| Stage | Count | % of Total | Many-to-One | Not Closest | Mean (m) | Median (m) | P95 (m) |
|---|---|---|---|---|---|---|---|
| Exact (UIC ref) | 21,226 | 33.3% | 13,199 | 5,786 | 20.56 | 8.51 | 77.42 |
| Name matching | 444 | 0.7% | 6 | 162 | 48.48 | 14.23 | 118.7 |
| Route matching (total) | 15,202 | 23.8% | 13 | 3,737 | 7.14 | 5.21 | 20.87 |
| GTFS | 15,202 | 23.8% | 13 | 3,737 | 7.14 | 5.21 | 20.87 |
| Distance (total) | 12,416 | 19.5% | 2,499 | 3,655 | - | - | - |
| Stage 0: Trio side matching | 2,462 | 3.9% | 2,462 | 861 | 8.55 | 4.46 | 23.2 |
| Stage 1: Group proximity (50m) | 8,982 | 14.1% | 29 | 2,362 | - | - | - |
| Stage 1a: UIC ref | 8,672 | 13.6% | 26 | 2,291 | 7.71 | 5.14 | 24.83 |
| Stage 1b: UIC name | 41 | 0.1% | 0 | 7 | 8.83 | 6.16 | 33.41 |
| Stage 1c: Name | 269 | 0.4% | 3 | 64 | 12.16 | 8.19 | 37.87 |
| Stage 1b: Group proximity (150m) | 490 | 0.8% | 3 | 367 | - | - | - |
| Stage 1d: UIC ref (150m) | 462 | 0.7% | 1 | 344 | 81.83 | 75.15 | 136.25 |
| Stage 1e: UIC name (150m) | 5 | 0.0% | 0 | 4 | 121.32 | 123.62 | 149.23 |
| Stage 1f: Name (150m) | 23 | 0.0% | 2 | 19 | 74.09 | 60.62 | 126.95 |
| Stage 2: Local ref | 73 | 0.1% | 4 | 25 | 16.21 | 12.89 | 39.9 |
| Stage 3a: Single candidate | 361 | 0.6% | 1 | 40 | - | - | - |
| 3a pass 1 | 326 | 0.5% | 1 | 34 | 12.68 | 8.59 | 44.18 |
| 3a pass 2 | 35 | 0.1% | 0 | 6 | 11.55 | 5.53 | 44.52 |
| Stage 3b: Ratio test | 48 | 0.1% | 0 | 0 | 2.86 | 2.42 | 6.62 |
| Post-processing: ATLAS Duplicate propagation | 334 | 0.5% | 334 | 123 | 24.16 | 8.69 | 101.53 |
| Post-processing: OSM stop partner propagation | 14,182 | 22.2% | 1,778 | 8,336 | 8.79 | 5.38 | 26.18 |
Distance Quality
Cross-Predicate Consistency
If we had matched every pair to its nearest OSM node instead:
Many-to-One Analysis
OSM platform/stop_position groups are counted as one OSM stop in this section.
Distribution: 1 ATLAS to N OSM stops
Ratio key: 1A:2O = one ATLAS platform matched to two OSM stops.
Distribution: N ATLAS to 1 OSM stop
Ratio key: 2A:1O = two ATLAS platforms matched to one OSM stop.
OSM Stops
OSM Stop Inventory
operator:wikidatanetwork:wikidataNon-node OSM Stops (Ways)
OSM Groups (Platform / Stop Position Pairs)
Groups by Type
| Group Type | Count |
|---|---|
| osm_trio | 1,232 |
| osm_pair_uic_equal_15m | 10,358 |
| osm_pair_uic | 2,944 |
| osm_pair_name | 653 |
| osm_pair_name_equal_15m | 30 |
| osm_pair_tram | 317 |
| osm_pair_tram_equal_15m | 12 |
Unmatched
Unmatched Analysis
Unmatched ATLAS (5,764)
| No OSM within 50m | 2,993 |
| Has nearby OSM | 2,771 |
Unmatched OSM (9,217)
| No ATLAS within 50m | 3,394 |
| Has nearby ATLAS | 5,823 |
| Has Attributes | Count |
|---|---|
| UIC & Local Ref & Routes | 170 |
| UIC & Local Ref | 102 |
| UIC & Routes | 2,196 |
| Local Ref & Routes | 87 |
| UIC only | 2,917 |
| Local Ref only | 86 |
| Routes only | 997 |
| None | 2,662 |
ATLAS
ATLAS Duplicates
| Total duplicate SLOIDs | 644 |
| Matched duplicates | 643 |
| Unmatched duplicates | 48 |
Problems
Problem Detection Statistics
| Problem Type | Count |
|---|---|
| Distance — Too far from OSM node | 1,214 |
| Unmatched — Isolated ATLAS / OSM node | 4,533 |
| Attributes — Missing or mismatched data | 925 |
| Problem Type | Count |
|---|---|
| Distance — Too far from OSM node | 4,544 |
| Unmatched — Isolated ATLAS / OSM node | 7,494 |
| Attributes — Missing or mismatched data | 338 |
| Duplicates — OSM or ATLAS side collisions | 342 |
| Problem Type | Count |
|---|---|
| Distance — Too far from OSM node | 5,623 |
| Unmatched — Isolated ATLAS / OSM node | 2,954 |
| Attributes — Missing or mismatched data | 10,973 |
| Duplicates — OSM or ATLAS side collisions | 1,453 |
Routes
Route Coverage
| ATLAS stops matched by GTFS routes | 43,770 / 55,009 | 79.6% |
| OSM stops with routes coverage | 35,235 | 71.3% |
Route-Route Matching
| Metric | Count |
|---|---|
| Route-route links (ATLAS ↔ OSM) | 2,059 |
| Linked ATLAS route IDs | 2,059 |
| Linked OSM route IDs | 2,059 |
| Coverage | Count | Percent |
|---|---|---|
| ATLAS route IDs with link | 2,059 / 4,028 | 51.1% |
| OSM route IDs with link | 2,059 / 2,874 | 71.6% |
| ATLAS itineraries with link | 1,056 / 14,345 | 7.4% |
| OSM itineraries with link | 1,056 / 8,206 | 12.9% |
| ATLAS route directions (total) | 7,599 | - |
| OSM route directions (total) | 4,920 | - |
Match Method Breakdown
| Method | Count | % of Total |
|---|---|---|
| Exact GTFS route ID | 1,663 | 80.8% |
| Normalized GTFS route ID | 396 | 19.2% |
| OSM display route ID matches ATLAS route ID | 0 | 0.0% |
Route Pipeline Shape
| Metric | Mean | Median |
|---|---|---|
| ATLAS itineraries per family | 3.56 | 3.0 |
| OSM itineraries per family | 2.86 | 2.0 |
| GTFS trips collapsed into one ATLAS itinerary | 122.75 | 17.0 |
| Matched family itinerary gap (ATLAS - OSM) | 0.99 | 0.0 |
| Matched Family Comparison | Count |
|---|---|
| Matched families where ATLAS has more itineraries | 899 |
| Matched families with equal itinerary counts | 810 |
| Matched families where OSM has more itineraries | 350 |
OSM Route Family Grouping Method
How OSM route relations are grouped into line families. Priority order: route_master → GTFS route ID → (route, ref, operator, network) → per-relation fallback.
| Grouping Method | Families | % of Families | Relations |
|---|---|---|---|
| route_master (explicit OSM grouping) | 2,672 | 93.0% | 7,959 |
| Normalized GTFS route ID | 5 | 0.2% | 7 |
| Synthetic key (route, ref, operator, network) | 197 | 6.9% | 240 |
| Per-relation fallback (no grouping key) | 0 | 0.0% | 0 |
| Total | 2,874 | 100% | 8,206 |
GTFS stop_id → ATLAS sloid Mapping
Matching Paths
① Direct SLOID match
Matches where original_stop_id directly contains a valid ATLAS SLOID.
② Strict matching
Matches on (UIC number, local ref) = (number, designation) in ATLAS.
③ Coordinate proximity
For remaining: ATLAS stops with the same UIC number within 0.5m (must be unambiguous).
④ Unique-number fallback
For remaining: if a UIC number maps to exactly one ATLAS row, use that sloid.
Cardinality
GTFS stop_id side
ATLAS sloid side
Unmatched Reasons
| Reason | Count |
|---|---|
| No ATLAS candidate for this UIC number | 687 |
| Non-unique ATLAS number (strict miss, fallback not applicable) | 32,034 |
| Coordinate proximity conflict (multiple candidates within radius) | 303 |