Background
Smukfest has an app where guests mark the artists they want to see. That creates a data source most festivals dream of — but nobody had used it for anything. The data was just sitting there.
I lead data and analytics at Smukfest, and I wanted to know: can we predict where people go? Not on average, but show by show, hour by hour. If 12,000 people walk from Shawn Mendes to Nik & Jay, security and logistics need to know in advance.
What I built
SmukFlow takes raw app data — 500,000+ favorites from nearly 25,000 guests — and calculates movement patterns between 194 shows across 7 stages over 8 days.
Tech stack
- SQLite for data with optimized indexes
- Python for computation — sequential path assignment that resolves time conflicts
- D3.js for interactive visualization of flows between stages
The algorithm builds a “path” for each guest based on their favorites, resolves overlaps via popularity, and aggregates all paths into crowd flows. The entire calculation runs in under 10 seconds.
The result is a map of the festival showing exactly which corridors get congested, when peak times hit, and how many people move between which stages.
Results
SmukFlow was built as a proof of concept on 2025 data and goes into production for 2026. The security team can now plan staffing based on data instead of gut feeling. Logistics knows where to place toilets, food stalls and bars. And festival management can see what happens when two popular artists play simultaneously on different stages.
