1. Collection
Rental and purchase listings, plus OpenStreetMap-derived business data, are collected over time.
Pipeline
From collection to public surfaces.
2. Normalization
Layout, area, age, station distance, address, and price fields are cleaned into comparable forms.
3. Spatial join
Addresses and town names are connected to wards, towns, station catchments, and polygons.
4. Analysis
Price structure, supply and demand, yield, and neighbourhood signals are modeled and visualized.
Models
Estimates are decision support, not appraisals.
OLS regression, LightGBM-style models, and structured aggregations help compare properties and areas. Outputs are not investment advice or formal valuation.
Normalize comparisons
Separate layout, area, age, station distance, and location effects where possible.
Read market movement
Use daily snapshots to track new supply, inventory, absorption, and price revisions.
Connect local activity
Use business density, category mix, and review growth as supporting local signals.
Caveats
Use the product with clear limits.
| Area | Limitation | How to use it |
|---|---|---|
| Geocoding | Some listings are placed at town level rather than exact building level. | Lean on town and ward comparisons rather than block-level precision. |
| Listings | Listings are not a complete transaction registry. | Read them as signals from the visible listing market. |
| Models | Estimates depend on input quality and available comparables. | Use them as comparison evidence, not the only basis for a decision. |
| Places | Review and listing behaviour differs by category. | Treat it as a supporting signal for how lively an area is. |