Skip to content

Pipeline

File Role
run_chain.py orchestrator: land-use step, activities, MATSim plans, MATSim, skims, loop
activities_to_matsim.py MAST output + land-use persons → MATSim plans
snap_to_network.py moves activities off car links outside the connected network
skims_from_matsim.py AM-peak car skims from MATSim events
transit_skim.py, TransitSkim.java transit skims with SwissRailRaptor
accessibility.py zone accessibility by year for the land-use models

Bookkeeping

Every stage writes <stage>/.done.json with its inputs, outputs, checksums, duration and settings, and is skipped next time unless --force is given. Completed stages are appended to registry.jsonl; logs go to pipeline/logs/. --dry-run prints the full command sequence.

Plans converter

python pipeline/activities_to_matsim.py --activities A --legs L --pp pp_2024.csv --hh hh_2024.csv \
       --out population.xml.gz [--expansion 10] [--residents-only] [--no-gateways] [--seed 20260707]

One weekday per person; tours cut to the Baltimore region (internal, inbound, outbound); homes on residential buildings and other activities on points of interest of the matching type in the zone.

Feedback loop

run_chain.py loop runs land use, activities and MATSim in two-year steps from 2016 to 2026 and feeds MATSim travel times back as skims, repeating a validation year until county population changes by less than 0.5 % and zone accessibility by less than 2 % (at most three repeats).