EXPLORE THE CITY

many layers.
one city.

POLICY SCENARIO

Intervention

Scenarios are evaluated on the server under the fitted posterior; choosing one reads a published result, it does not rerun the model.

VIEW

customize layers

how this works

936,597 people are drawn from the census and the commuter statistics, each with a home cell, a workplace and a survey diary.
Every shop trip picks a venue by measured floor area and distance; three demand classes (food, clothing, long-term goods) have their own kernels.
Each visit spends a household's share of the district's purchasing power; leakage and in-commuter spend come from the published class balances.
The kernel parameters are inferred from the 69 published district × class turnovers. The result is a set of plausible kernels, 48 short MCMC chains: an approximate posterior, not a calibrated one.
A policy scenario re-runs the same people and money under a changed venue set, under every posterior draw. Its numbers are model results under stated assumptions, not forecasts.

the model, with every definition named in the code ↗

City

Region to explore

Choose a district here or click its boundary on the map. A district is required for observed record overlays. Selecting one filters existing data and published outputs; it does not run a new simulation.

Observed storefront / licence year

Year filters source records. It does not change the date of a published simulation.

Population unitsresidents per approximately 100 m cellColor is clipped at 400 residents/cell. For context, 100 residents/cell ≈ 10,000 residents/km².

3D column height 1×

lowerhigher

Changes vertical exaggeration only. Population values and colors stay fixed.

Data connections

Evidence APISimulation API

STUTTGART / WHOLE CITYobserve. simulate. compare.

Columns show population, not building height. Lines show model choices, not observed journeys.

Time08:00

Playback speedTrail

Playback reads a recorded weekday; it does not advance the model.

Loading the published-data fixture…