The author created maps showing walking routes for people in Brooklyn to their nearest L train stop, using data from the US Census and LODES dataset. They used Valhalla's routing solution to generate pedestrian routes, but found that with many overlapping segments, it was difficult to gauge the number of people along each route. To address this issue, they developed a PL/PGSQL function to segment the lines into straight parts with no turns, and then aggregated the data to style the routes by line thickness and color based on the number of commuters and maximum commute time along each segment. The final map shows powerful and impactful visualizations of the walking commutes, allowing users to better understand the impact of the L train closure on local communities.