Using Causal Inference to Estimate the Impact of Tube Strikes on Cycling Usage in London

A walkthrough of applying causal inference to estimate the effect of London tube strikes on Boris/santander bike usage. The author processes 144 weekly TfL CSVs into a panel dataset at H3 cell-day level, defines treatment as proximity to striking tube lines, and estimates a ~3.95% increase in bike trips on strike days using a Two-Way Fixed Effects (twfe) regression with clustered standard errors. The post covers the potential outcomes framework, ATE vs ATT, selection bias, parallel trends, positivity, no-anticipation, and sutva assumptions, while progressively refining the dataset to improve signal-to-noise ratio by focusing on central interchange stations and days near strike events.