Taxi demand by city and time

Choose a country and a city

Pick a country, then a city, to see its estimated taxi demand on a map by area, weekday and half-hour. The link in your address bar keeps your choice, so you can share it.

Compare cities

All 32 cities side by side, grouped by country. Click a column title to sort; click a city to open it on the map. All numbers are model estimates.

What the cities have in common

The city-wide weekly rhythm is the same in every city; what differs is where in the city demand falls. These findings are computed from the model estimates of all published cities.

• The busiest half-hour is the same everywhere

  In all 32 cities, the estimated busiest half-hour of the week is Thursday from 2:30 PM (14:30). The weekly timing is learned from taxi trips in Chicago, New York and San Francisco and is not estimated city by city; per half-hour, weekday mornings from 7 to 9 AM have about 1.2 times the pickups of Friday and Saturday nights from 10 PM to 2 AM.

• How concentrated demand is

  The ten busiest areas of a city hold a median of about 43% of its estimated demand. The most concentrated cities are Glasgow (65%), Leeds (55%), Sheffield (55%), Melbourne (54%) and Cardiff (54%).

• Airports

  13 of the 32 cities have a large airport inside the study area. The airport area's share of the estimate is largest in Miami (24%), Los Angeles (23%), Boston (20%), Coventry (19%) and Manchester (15%). In 9 of them the airport area is the single busiest area of the year.

• City centres

  The areas within 2.5 km of the city centre hold a median of about 28% of a city's estimated demand (32 cities).

• Weeknight mornings and weekend nights

  In 3 of the 32 cities, the busiest area on Friday at 10 PM is a different area from the busiest on Tuesday at 8 AM, so where demand falls changes through the week even though the city-wide timing does not.

How the estimate is made

A research model learned from one year of taxi trips in Chicago, New York and San Francisco how taxi pickups relate to what is on the map: restaurants and bars, shops, offices, stations and bus stops, schools, hospitals and clinics, leisure and tourist places, and the airport. It never sees any taxi data of the cities on this page.

For each city, the model splits the city into small areas (suburbs, ZIP code areas, wards or postal-code areas), counts the public OpenStreetMap places in each, and adds the calendar and the size of the airport. From these it estimates each area's share of the city's taxi pickups for every weekday and half-hour. Twenty model runs are averaged where available; the number of runs is shown for every city.

The weekly rhythm of the whole city, which hours and days are busy, is the average of the three training cities, so it is the same for every city. The share of each area changes by weekday and time, and that is what the map shows. This approach is called strict zero-shot transfer.

What it cannot tell you

• It does not give real counts of pickups. Every number is a model estimate.
• It has not been checked against local taxi trips in these cities.
• It describes taxi-style pickups as recorded in the training cities, not rideshare apps.
• The yearly number of trips in a city is not known, so the map shows shares and a demand index by default.
• It shows a typical week of the calendar year, not a specific day.
• It includes no weather, events, roadworks or current traffic.

Questions and answers

Data, code and paper

The estimates are free to download and use under the CC BY 4.0 licence. Please credit Miliroute and the authors, and say that the numbers are model estimates.

• List of all cities (JSON) · API description (OpenAPI) · All key facts as plain text
• Code: github.com/MLDSPARROW/strict-zero-shot-taxi-demand
• Paper: Ghalejughi, M., Khattak, H. A., Hussain, Z. Spatial-scale-aware cross-city transfer for zero-shot taxi demand estimation (in preparation).

Files for Sydney, NSW, Australia: JSON · areas CSV · week CSV (gzip). Choose another city above to get its files; the city list has every file address.

Cite this page

Ghalejughi, M., Khattak, H. A., Hussain, Z. (2026). Miliroute taxi demand estimates, data version v1-2026-10-25. Model estimates by strict zero-shot transfer; not checked against real trips. https://miliroute.com/taxi-demand/

@misc{miliroute_taxi_demand_2026,
  author = {Ghalejughi, Milad and Khattak, H. A. and Hussain, Z.},
  title = {Miliroute taxi demand estimates},
  year = {2026},
  version = {v1-2026-10-25},
  howpublished = {\url{https://miliroute.com/taxi-demand/}},
  note = {Model estimates by strict zero-shot transfer, not checked against real trips. Method: Spatial-scale-aware cross-city transfer for zero-shot taxi demand estimation (in preparation)}
}

Copy citation Copy BibTeX

Sources and licences

• Map data © OpenStreetMap contributors (ODbL)
• Contains public sector information licensed under the Open Government Licence v3.0. Wards (December 2024) Boundaries UK BGC: Office for National Statistics; contains OS data © Crown copyright and database right 2024; airport passengers: UK Civil Aviation Authority, 2019
• Suburb outlines: Geoscape Administrative Boundaries (CC BY 4.0). Jobs and residents in 2021: Transport for NSW Travel Zone Projections (CC BY). Commuters: ABS Census 2021 (CC BY 4.0). Airport boardings: half of the 2019 passengers of Sydney Airport
• Suburb/locality boundaries: Geoscape Administrative Boundaries (Geoscape Australia and the Department of Industry, Science and Resources), CC BY 4.0; airport passengers: BITRE and airport reports, 2019
• US Census Bureau, ZIP Code Tabulation Areas 2020, cartographic boundary file (public domain); airport boardings: FAA calendar year 2019 enplanements (public domain)

Change log

• 2026-10-25: Added 18 cities: Belfast (United Kingdom); Boston (United States); Brighton (United Kingdom); Bristol (United Kingdom); Canberra (Australia); Cardiff (United Kingdom); Coventry (United Kingdom); Denver (United States); Leicester (United Kingdom); Miami (United States); Minneapolis (United States); Newcastle (Australia); Newcastle upon Tyne (United Kingdom); Nottingham (United Kingdom); Pittsburgh (United States); Portsmouth (United Kingdom); Southampton (United Kingdom); Wollongong (Australia).
• 2026-10-24: Added 14 cities: Adelaide (Australia); Birmingham (United Kingdom); Brisbane (Australia); Glasgow (United Kingdom); Leeds (United Kingdom); Liverpool (United Kingdom); London (United Kingdom); Los Angeles (United States); Manchester (United Kingdom); Melbourne (Australia); Perth (Australia); Philadelphia (United States); Sheffield (United Kingdom); Sydney (Australia).