Every dot in this dataset is a home. Every arc is a commute. Built from 8,707 real origin → destination job flows across San Francisco, this is an interactive geospatial intelligence explorer of urban commute patterns — where they concentrate, which direction they favor, and how far people actually travel to work.
Filtered to San Francisco home → work pairs with valid coordinates. Every figure below is computed directly from the underlying commute records.
Every arc represents movement from home to work. Arc width represents flow magnitude — the number of jobs (S000) making that commute.
Reset returns the map to its default bearing/pitch. Fullscreen expands the frame for detailed inspection. Pan, zoom and rotation are native to the embedded deck.gl scene — drag to orbit, scroll to zoom.
1. Each arc represents an origin → destination relationship: a home location connected to a work location.
2. Arc width represents S000 — the job count on that commute.
3. Orange marks the source (home); green marks the target (work).
4. Hover any arc for its exact job count. Pitch and bearing create the 3D perspective — drag to rotate.
Explore the city from different angles.
Reveal local flow concentrations.
Inspect individual commute flows.
Look for dominant movement corridors.
The Commute Compass abstracts away geography entirely. Each wedge is a compass bearing computed from every home → work pair; its length is total jobs, its colour is average commute distance.
The eastern bearing carries the largest aggregated flow in the filtered dataset — 14.2% of all 23,467 jobs. Commutes heading north-east travel farthest on average (2.05 km), a pattern the 3D arc map hints at but never states directly.
Jobs are not evenly distributed across bearings — the strongest direction (E, 3,337 jobs) carries roughly 5.5× the flow of the weakest (SW, 605 jobs). Movement concentrates east and east-south-east of the filtered home locations, tapering sharply toward the south-west.
69% of jobs in this filtered area involve a straight-line commute of under 2 km — a dense, walkable catchment rather than a long-haul commuter zone. Distances beyond 4 km account for less than 0.4% of jobs.
The ten strongest home → work relationships in the dataset, ranked by job count. Most pairs move 2–3 jobs; these ten alone account for 178 jobs — 0.76% of the total — on just ten specific routes.
Four data-driven readings of the same underlying dataset — expand each for the numbers behind it.
Flows cluster tightly around a compact eastern corridor of the study area.
Most pairs carry 2–3 jobs; a handful of corridors carry 16–21.
69% of jobs travel under 2 km — short, walkable commutes dominate.
East carries 5.5× the flow of the weakest bearing, south-west.
The official workflow required validation, cleaning, spatial filtering, ArcLayer creation, a 3D view, tooltip interaction, and static HTML export. That pipeline is preserved end-to-end below.
lat_h, lng_h, lat_w, lng_w) and the flow field (S000) checked for completeness and numeric validity.ArcLayer renders each pair with getWidth = S000 × 2, orange source colour, green target colour, and a dark-matter basemap.lng_hHome longitudelat_hHome latitudelng_wWork longitudelat_wWork latitudeS000Flow / job count — number of jobs on that home → work relationship