mbta-performance¶

Turns the MBTA's LAMP files into the per-stop event CSVs the dashboard reads. Also builds travel-time benchmarks and alert files.

Repo


Stack	Python Chalice · pandas + pyarrow
Runs on	Lambda (stack `mbta-performance`)
Deploys	Push to `main`

flowchart LR
    lamp(["LAMP<br/>subway, bus, alerts"]) --> perf["mbta-performance<br/><small>every 30 min + daily</small>"]
    gtfs[("tm-gtfs")] --> perf
    szarch[("SlowZones/ archive")] --> perf
    perf --> events[("tm-mbta-performance<br/>Events-lamp/, Alerts/lamp/,<br/>Benchmarks-tm/")]
    perf -. missing feed .-> sqs[["SQS → data-ingestion"]]

Good to know¶

Today's subway data is processed every 30 minutes. Yesterday's subway and bus data is reprocessed once a day.
Historical monthly files are loaded by hand. Follow the historic runbook.

lamp¶

transitmatters/lamp is our fork of mbta/lamp, the MBTA's own pipeline. We don't run it. We use the fork to send improvements upstream so the public LAMP data has what we need.

Naming trap

Inside LAMP, tm means TransitMaster (the MBTA's bus system), not TransitMatters.