WP2·Mobility and physical activity synthetic

A city on the move, made up.

200 synthetic people, a week of GPS every 30 seconds, built to the protocol's numbers and run through its cleaning. No one was tracked.

Watch a day ↓
07:00average weekday · synthetic
Synthetic data
01 / 05

One phone, one day.

A fix every 30 seconds. Indoors, most of them are noise.

–raw fixes, one synthetic person-day
02 / 05

Cleaned by the protocol's rules.

Fixes worse than 5 m go, impossible jumps go, short gaps are filled. Each fix gets a mode.

166,104of 663,809 fixes pass the 5 m rule
03 / 05

Everyone at once, through the day.

Minutes of walking and cycling per 250 m square, every half-hour.

0walking
0cycling
04 / 05

Which mode was it?

Scored against the truth the generator kept. Walking is hardest: 14% of walking fixes are read as stays, 10% as cycling.

05 / 05

What they walk past.

Each walking or cycling fix takes the green view of the street within 25 m, from WP3.

r 0.96observed vs true half-hour green view
Not a finding

The rhythm is the one the generator was given.

Commutes and lunch walks were written into the generator. The run shows that the protocol's chain works end to end and can be scored against a known answer. It says nothing about how Amsterdam moves.

The scores flatter the chain, because synthetic speeds are more regular than real ones. Where real data will bite: the 5 m rule removes most indoor fixes, buses are hard to tell from bicycles by speed alone, and exposure is missing wherever a path runs along a street the image sample did not cover.

Details
What the protocol fixes, and what is assumed here

From the protocol: GPS every 30 s, steps, seven days, the 5 m accuracy rule, the speed rule, gap filling up to 2 minutes, stay points, 25 m street exposure, half-hours. Assumed here, labelled and not calibrated: a 07:00–19:00 window, homes across CBS income strata, OpenStreetMap destinations with distance decay, mode by trip distance (walks over 3 km become cycling), ferries, GPS error indoors and out, signal gaps, non-wear.

Steps were calibrated to the protocol's figures: person means average 9,587 a day (SD 2,248), against 9,448 (SD 2,218).

The spike map

Every recorded fix of a synthetic person walking or cycling, by its true state, in 250 m squares and half-hours from 07:00 to 19:00, as minutes per average weekday (five days) or weekend day (two days). People counts are the synthetic people walking or cycling in that half-hour on an average day.