# Open Dubai Property Graph: crosswalk match report (as_of 2026-10-26)

Window: 12 months to as_of (sales by instance_date, Ejari by contract_start_date). Rows: archetypes 25,980, crosswalk 22,668.

| Source (12 months) | n | building | project only | transaction name only | unassigned |
|---|---:|---:|---:|---:|---:|
| DLD sales · all | 151,182 | 50.0 % | 44.4 % | 4.9 % | 0.7 % |
| DLD sales · apartment offplan | 107,055 | 58.7 % | 41.1 % | 0.2 % | 0.0 % |
| DLD sales · apartment ready | 28,240 | 44.9 % | 29.4 % | 25.7 % | 0.0 % |
| DLD sales · villa offplan | 10,583 | 0.0 % | 100.0 % | 0.0 % | 0.0 % |
| DLD sales · villa ready | 5,304 | 0.0 % | 79.6 % | 0.0 % | 20.4 % |
| Ejari · all | 613,449 | 15.6 % | 15.9 % | 0.0 % | 68.5 % |
| Ejari · apartment new | 218,778 | 28.3 % | 17.5 % | 0.0 % | 54.2 % |
| Ejari · apartment renewal | 331,676 | 10.1 % | 8.4 % | 0.0 % | 81.5 % |
| Ejari · villa new | 28,706 | 0.0 % | 64.7 % | 0.0 % | 35.3 % |
| Ejari · villa renewal | 34,289 | 0.0 % | 38.0 % | 0.0 % | 62.0 % |

## What the columns mean

- **building**: assigned to one DLD registry building (its ID in `units`): the project has a single residential building, or the building has no project
- **project**: assigned to a DLD project only: the project has several buildings (the registry does not name them, so a tower cannot be told apart) or it is a villa project
- **transactions_only**: sale with no DLD project: kept under its own building name within the DLD area, with no registry stock and no Ejari
- **unassigned**: no project and no building name (Ejari never names the building)

## How records are matched

project_number: the DLD project number in the record matches the `projects` table (official ID). project_name: no usable number, but the exact project name (case and punctuation ignored) matches exactly one project_id in `units` or `buildings` in the same DLD area. Nothing else is matched: no fuzzy names, no guessing which tower is which.

- Sales in 12 months by method: {'project_number': 142379, 'none': 8563, 'project_name': 240}
- Ejari contracts in 12 months by method: {'none': 420172, 'project_number': 191519, 'project_name': 1758}
- Check: 300,208 sales carry both a project number and a matching project name; 0 disagree. Ejari: 370,005 and 0.

## Registry

- residential units: 1491416
- units assigned pct: 82.0
- units without project pct: 48.4
- units without building or project: 268021
- units without bedrooms: 42310
- villas: 225072
- villas without project: 78856
- registry rows assigned pct: 79.8

## Where the crosswalk fails

- Ejari (rent_contracts) never names the building, and about two thirds of contracts carry no DLD project, so they can only be linked to an area or master project. Rents are attached to a building only when the project has one residential building.
- In projects with several towers, `units` identifies each building by an ID but not by name, and `transactions` names the tower but gives no ID. Those projects are published at project level, never split by guesswork.
- Older freehold buildings with no DLD project in the records (International City, Discovery Gardens, The Greens, parts of Business Bay and others) appear only under their transaction building name: no registry stock and no rent.
- The units registry includes off-plan units registered before completion, so stock and turnover in projects under construction count units that are not built yet.
- Bedrooms come from the DLD record; sales and units without a bedroom count are left out of the archetypes. Villa sales without bedrooms take them from the buildings registry by exact plot size (same rule as the site).
- The `projects` table is the version served by data.dubai on the snapshot date (see datapackage sources), which may be newer than as_of.

DLD areas with the most sales not linked to a registry building (12 months): Jabal Ali First (1,068), Marsa Dubai (1,052), Al Warsan First (932), Wadi Al Safa 3 (526), Al Hebiah First (504), Madinat Al Mataar (438), Al Thanyah Third (391), Al Barsha South Fourth (390), Burj Khalifa (307), Wadi Al Safa 6 (253), Al Thanyah Fifth (252), Warsan Fourth (241)

DLD areas with the most unassigned Ejari contracts (12 months): Jabal Ali First (21,732), Al Warsan First (21,215), Al Nahda Second (15,958), Al Barsha First (14,325), Al Karama (12,784), Al Murqabat (12,140), Muhaisanah Fourth (11,995), Mirdif (11,727), Al Nahda First (11,602), AL QUSAIS (11,418), Mankhool (11,231), Nadd Hessa (10,915)

Source: Dubai Land Department (DLD) open data, published on data.dubai. Analysis: FairSqft. Not endorsed by the DLD or the Government of Dubai.
