Mobi Transit Explorer: A Reproducible Pipeline and Interactive Analysis of Nine and a Half Years of Vancouver Bike-Share Trips (2017–2026)

Public bike-share programs publish rich trip archives, but the raw data is messy in ways thatdefeat casual analysis: schemas drift year to year, identifiers change format, and sentinel valuesmasquerade as measurements. This report describes Mobi Transit Explorer, an end-to-end sys-tem that turns nine and a half years (2017–2026) of Vancouver’s Mobi bike-share trip files intoa reproducible analytical warehouse and a public, interactive explanation of how the networkfunctions as an extension of transit. The central engineering problem is one of data integra-tion: reconciling 103 heterogeneous source files spanning 31 distinct column layouts into a singlecanonical schema without silently guessing. Over the warehouse we define transparent, docu-mented metrics—a transit-connector score, a leisure heuristic, rule-based opportunity findings, acorroborated wildfire-smoke-day flag, and per-station reported-crash context—and expose themthrough a browser interface built for non-expert readers. We report network-level findings(growth from 547,355 annual trips to a 2023 peak of 1,234,500, a 35.6% pandemic contractionand recovery, e-bikes rising from first appearance to 40.9% of trips in the first half of 2026,a structural shift toward institutional memberships, ridership that shows no obvious responseto typical wildfire smoke though the one sustained severe episode cut it sharply, and a binarypattern of dock coverage at rapid-transit stations) and document the limitations candidly. Thecontribution is the assembly and its discipline: a catalogue of real-world schema drift, reconcilia-tion that stops rather than guesses, and report prose held to the artifacts by machine verification.Its open questions—automated reconciliation of drifting civic datasets, and trustworthy data ex-ploration for non-experts—sit in the data-management and human–data-interaction traditions.