Event-impact forecaster

The economics of live events.

Own a business in San Francisco? See how much foot traffic a Giants home game sends past your door, block by block.

Game-day lift at your business

Search by name or address, or click the map to drop a pin.

Any date through the end of the current schedule.

Pick your business above or click your spot on the map.

Click your spot, or search on the left

How it works

Two counterfactual models learn what every city block looks like on a normal day, trained only on hours with no event: a gradient-boosted benchmark that scores each block on its own, and a graph network that lets related blocks inform each other. Each applies its own measured game-day effect on top, band by band, out from the ballpark. Pick your spot and a date; the schedule fills in the rest, and the toggle shows where the two models agree and where the estimate leans on model structure. Numbers come from the live models, in visitor-hours over the 4pm to 11pm window: one visitor-hour is one person present for one hour, so it is not a headcount.

Built on the USD AAI-590 capstone by Stephen Farmer, Johnathan Kelsey, and Lucas Young. Foot-traffic data: Advan Research. Basemap: CARTO, OpenStreetMap contributors.