Location intelligence co-location archetypes

@@ -29,8 +29,8 @@ The three-letter codes were ratified on 1 June 2026.
| Code | Name | Anchor type | Status |
|------|------|-------------|--------|
| **PRO** | Retail Centres | Grocery hypermarket with hardware and at least one of: price club, lifestyle, or electronics | Live — T1/T2/T3 co-location pipeline |
| **VWH** | Urban Fringe | Hardware + trade-supply ecosystem (MRO, tool rental, builders merchant, auto parts) | Live — 6,368 clusters (T1=852 / T2=1,327 / T3=4,189) |
| **PKS** | Commuter | Regional transit anchor (airport, rail, bus) + park-and-ride + car rental/hotel enrichment | Live — 6,953 clusters (T1=691 / T2=2,658 / T3=3,604) |
| **VWH** | Urban Fringe | Hardware + trade-supply ecosystem (MRO, tool rental, builders merchant, auto parts) | Live — production co-location pipeline across three tiers |
| **PKS** | Commuter | Regional transit anchor (airport, rail, bus) + park-and-ride + car rental/hotel enrichment | Live — production co-location pipeline across three tiers |

PRO is the base map product — the foundation of the site-selection dataset.
VWH and PKS are overlay archetypes that identify adjacent market structures
@@ -50,25 +50,20 @@ on anchor composition.
**T1 — Regional:** A cluster containing a grocery hypermarket and a hardware
retailer, plus at least one of a price club, lifestyle retailer, or electronics
retailer. Alternatively: four or more anchor-category retailers in a tight
cluster (span ≤ 1 km), or three or more anchors in any tight cluster.
cluster, or three or more anchors in any tight cluster.

**T2 — District:** A cluster containing a grocery hypermarket and a hardware
retailer, with a span no greater than 2.5 km.
retailer, within a wider district-level span.

**T3 — Local:** All remaining anchor pairs that do not qualify for T1 or T2.

### Current dataset (Phase 23 + Change B, 28 May 2026)
### Current dataset

| Tier | Clusters | Countries |
|------|----------|-----------|
| T1 | 1,746 | 17 |
| T2 | 2,726 | 17 |
| T3 | 2,021 | 17 |
| **Total** | **6,493** | |

The dataset covers 17 display countries across North America and Europe.
The T2 span boundary was set to 2.5 km in the Change B rebuild, tightening
the T2 boundary relative to earlier phases.
The current production dataset spans thousands of PRO clusters across all
three tiers, covering 17 display countries across North America and Europe.
Tier boundaries are periodically re-tuned as the underlying retail footprint
changes; the T2 district radius has been tightened in successive rebuilds
relative to earlier phases.

---

@@ -76,15 +71,15 @@ the T2 boundary relative to earlier phases.

VWH clusters identify concentrations of hardware and industrial-supply
retailers in the absence of grocery anchors. These sites occupy the urban
fringe — locations between 5 and 80 km from a major metro centre — and
tend to cluster around highway interchanges in areas with adjacent industrial
landuse.
fringe — a band of distance beyond the immediate metro core but short of
standalone-market territory — and tend to cluster around highway interchanges
in areas with adjacent industrial landuse.

### Definition

A VWH candidate is a location where one or more hardware retailers are
present, no grocery hypermarket is within the cluster span, and the site
sits within metro-distance 5–80 km. The typical built form is a 3–6 story
sits within the Urban Fringe metro-distance band. The typical built form is a
multi-storey warehouse or light-manufacturing building, distinct from the
one-storey big-box format of the retail park.

@@ -97,46 +92,41 @@ and just-in-time logistics tenants — not general retail consumers.

| Signal | Rationale |
|--------|-----------|
| Highway interchange ≤ 2 km | Truck ingress and egress |
| Population ≥ 300,000 within 30-minute drive | Manufacturing and logistics labour |
| Highway interchange nearby | Truck ingress and egress |
| Sufficient population within a short drive | Manufacturing and logistics labour |
| Industrial landuse adjacent | Zoning compatibility |

**Significant:**

| Signal | Rationale |
|--------|-----------|
| Air cargo airport ≤ 20 km | Electronics and components, rapid replenishment |
| Freight rail ≤ 2 km | Just-in-time component delivery |
| Transit corridor ≤ 500 m | Workforce access |
| Air cargo airport within reach | Electronics and components, rapid replenishment |
| Freight rail nearby | Just-in-time component delivery |
| Transit corridor nearby | Workforce access |

**Disqualifying:** Dense residential immediately adjacent; flood plain;
heritage conservation zone; location inside a PRO cluster.

### Production results (11 June 2026)

The VWH pipeline is production-grade. Hardware stores (10,338 locations, 45 chains)
were profiled as proxy anchors; DBSCAN was run on trade-supply POIs without the hardware
anchor (held-out validation); tier rules use group-collapse logic validated at 73.4%
hardware co-location on T1+T2 clusters (acceptance threshold: 55%).

| Country | Clusters |
|---------|---------|
| United States | 3,167 |
| Germany | 648 |
| United Kingdom | 543 |
| Canada | 506 |
| France | 420 |
| Netherlands | 240 |
| Italy | 226 |
| Poland | 171 |
| **Total (17 countries)** | **6,368** |

Tier distribution: T1 (Full Trade Hub) = 852 (13.4%), T2 (Established) = 1,327 (20.8%),
T3 (Emerging / Thin) = 4,189 (65.8%). T3-heavy distribution is expected: full trade hubs
combining MRO, tool rental, builders merchant, and auto parts are legitimately rare.

A `retail_contamination` flag marks clusters where a grocery hypermarket lies within 1 km
of the centroid (3,048 clusters; 47.9%). These are dual-use commercial parks — valid VWH
### Production status

The VWH pipeline is production-grade. Hardware stores were profiled as
proxy anchors, and the trade-supply clustering was validated against
held-out hardware-anchor data, with an internally set acceptance threshold
for cluster quality that the production build clears.

The dataset spans thousands of clusters across the 17 display countries,
concentrated most heavily in the United States with meaningful coverage
across several other North American and European markets.

Clusters are distributed across the three tiers in the expected shape: a
small minority reach the Full Trade Hub tier, a larger share reach the
Established tier, and the majority sit in the Emerging/Thin tier. That
T3-heavy distribution is expected — a full trade hub combining MRO, tool
rental, builders merchant, and auto parts is a legitimately rare
combination.

A quality-control flag marks clusters that sit close enough to a grocery
hypermarket to be considered dual-use commercial parks — valid VWH
co-locations that also include grocery retail.

---
@@ -144,17 +134,17 @@ co-locations that also include grocery retail.
## PKS — Commuter

PKS clusters identify commercial concentrations near regional airports and
intercity train stations that sit in a Commuter belt 15–150 km from a major
metro centre. The defining demand pattern is park-and-fly or park-and-train
travel: residents of a Regional Market drive to a transit node, park, and
travel to the Metro Market.
intercity train stations that sit in a Commuter belt beyond the immediate
metro core but short of standalone-market distance. The defining demand
pattern is park-and-fly or park-and-train travel: residents of a Regional
Market drive to a transit node, park, and travel to the Metro Market.

### Definition

A PKS candidate is a regional transit node — airport or intercity train
station — at metro distance 15–150 km. Nodes within 15 km of the metro
centre are classified as suburban rather than regional; nodes beyond 150 km
are considered standalone markets with a separate metro relationship.
station — within the Commuter metro-distance band. Nodes closer than that
band are classified as suburban rather than regional; nodes beyond it are
considered standalone markets with a separate metro relationship.

The defining commercial signal at a PKS location is car rental. Auto parts,
fuel stations, quick-service restaurants, and convenience stores are
@@ -166,52 +156,52 @@ secondary signals.

| Signal | Rationale |
|--------|-----------|
| Regional transit anchor ≤ 3 km | Airport or intercity station with direct metro service |
| Metro isolation 15–150 km | Defines the regional relationship |
| T1 or T2 cluster ≤ 10 km | Same population generates parking demand |
| Regional population ≥ 150,000 | Minimum demand for multi-storey parking |
| Regional transit anchor nearby | Airport or intercity station with direct metro service |
| Metro isolation within the Commuter band | Defines the regional relationship |
| PRO T1 or T2 cluster nearby | Same population generates parking demand |
| Sufficient regional population | Minimum demand for multi-storey parking |

**Significant:**

| Signal | Rationale |
|--------|-----------|
| Car rental ≤ 1 km | Arriving travellers require transport |
| Hotel cluster ≤ 500 m | Business travel and multi-day parking |
| Second transit mode ≤ 5 km | Multi-modal integration |
| Car rental nearby | Arriving travellers require transport |
| Hotel cluster nearby | Business travel and multi-day parking |
| Second transit mode nearby | Multi-modal integration |

**Disqualifying:** Major hub within 15 km; population under 100,000; no
direct metro service.
**Disqualifying:** Major hub within the immediate metro core; population
below a minimum viable threshold; no direct metro service.

### Production results (11 June 2026)
### Production status

The PKS pipeline is production-grade. Park-and-ride records (23,117 locations) serve
as the primary geographic anchor — actual car→transit transition points distributed
independently of rail network geometry. Transit modes are enrichment signals; car rental
and hotel presence define commercial maturity. Tier rules use mode-group collapse
(intercity_rail + commuter_rail collapse to the RAIL group, preventing artificial
bimodality inflation).
The PKS pipeline is production-grade. Park-and-ride records serve as the
primary geographic anchor — actual car-to-transit transition points
distributed independently of rail network geometry. Transit modes are
enrichment signals; car rental and hotel presence define commercial
maturity. Related transit-mode categories are grouped together before
tiering, to prevent related modes from inflating an apparent multi-modal
signal.

| Tier | Clusters | % | Definition |
|------|---------|---|-----------|
| T1 (Regional Hub) | 691 | 9.9% | Multi-modal + full commercial ecosystem |
| T2 (Transit Interchange) | 2,658 | 38.2% | Transit + at least one commercial signal |
| T3 (Transit Node) | 3,604 | 51.9% | Transit present; commercial opportunity |
| **Total** | **6,953** | | |
Clusters are distributed across three tiers. A Regional Hub tier combines
multi-modal access with a full commercial ecosystem. A Transit Interchange
tier combines transit with at least one commercial signal. A larger Transit
Node tier is where transit is present but commercial opportunity remains to
be proven out.

Commercial enrichment: car rental chains (hertz-eu, avis-eu, europcar-eu, sixt-eu, and
others) and hotel chains (ibis-eu, premier-inn-gb, holiday-inn-express-us, and others)
are all ingested and active in the production build.
Commercial enrichment draws on major car rental and hotel chains active in
each market, ingested and reflected in the production build.

### Major hub filter

Airports with a T1 PRO cluster within 5 km are excluded as likely major commercial hubs.
Major airports generate their own retail gravity and do not exhibit the park-and-transit
pattern. The filter correctly removes LAX, JFK, LHR, and CDG.
Airports adjacent to a major PRO retail cluster are excluded as likely major
commercial hubs. Major airports generate their own retail gravity and do not
exhibit the park-and-transit pattern. The filter correctly removes hubs such
as LAX, JFK, LHR, and CDG.

### Future enhancements

- Airport passenger volume data (CAPA or IATA) to replace the T1-adjacency hub proxy
  with a direct traffic-based classifier
- Airport passenger volume data (CAPA or IATA) to replace the current
  adjacency-based hub proxy with a direct traffic-based classifier
- Parking operator directory: Q-Park, APCOA, NCP, Indigo/Vinci (EU); SP+ (US)

---
@@ -222,10 +212,10 @@ VWH and PKS appear as overlay layers under the **★ Regional Markets** section
in the layer control panel.

**VWH toggle** — displays orange dots at Urban Fringe candidate locations.
When active, cluster bubbles fade to 10% opacity to reduce visual interference.
When active, cluster bubbles fade to reduce visual interference.

**PKS toggle** — displays teal dots at integrated candidates (T1/T2 cluster
within 10 km) and grey dots at standalone candidates. The same 10% bubble
**PKS toggle** — displays teal dots at integrated candidates (near a PRO
T1/T2 cluster) and grey dots at standalone candidates. The same bubble
fade applies.

Both layers persist across view transitions — the fade state is maintained