How It Works

Crowd Score Calculation

The crowd score (0–100) is computed from live posted wait times for ~15 headline attractions per park (15 at Disneyland, 17 at DCA). We take the median wait time of operating headliners and map it to a score using a piecewise function: • 0–20 → Light (median ≤ 18 min) • 21–40 → Moderate (median 19–36 min) • 41–60 → Busy (median 37–54 min) • 61–80 → Packed (median 55–72 min) • 81–100 → Slammed (median 73+ min) Attractions that are closed or down are excluded from the median calculation to avoid skewing the score. If fewer than 3 headliners are reporting, the score is withheld until more data is available.

Data Sources & Freshness

Wait times are sourced from Queue-Times.com and refreshed every 5 minutes via server-side caching. Snapshots are persisted to our database every 15 minutes, building a growing historical dataset used for prediction calibration. Data pipeline: • Queue-Times API → Edge function (5-min cache) → Live display • pg_cron job → 15-min snapshots → wait_snapshots table • Nightly batch (3 AM UTC) → 12-month forecast regeneration The "last updated" timestamp reflects the most recent successful data fetch from the upstream provider. Provider health is monitored continuously and surfaced on the live pages.

Prediction Method (v1 Heuristic)

The predictions use a transparent, rules-based heuristic model. No black box — here's exactly what goes in: Input signals (weighted): • Day-of-week patterns — weekends average 15–25% higher crowds • Week-of-year seasonality — mapped to known peak/valley periods • US federal holidays + adjacent weekends (e.g., 4th of July week) • SoCal school district break windows (LAUSD, OUSD calendars) • Disney-specific events: Lunar New Year, Food & Wine Festival, Oogie Boogie Bash, Grad Nites, holiday overlays • Weighted average of similar historical dates from our snapshot archive Confidence levels: • High — date is within 2 weeks; strong historical coverage for similar day/week patterns • Medium — 2–6 weeks out; moderate data, some event uncertainty • Low — 6+ weeks out; limited historical match, event schedule may not be finalized Model updates: Predictions are regenerated nightly at 3 AM UTC. When Disney announces new events, ticket deals, or schedule changes, we update the event calendar and factors manually — typically within 24–48 hours of announcement. This is a v1 heuristic model; future versions will layer in machine learning for improved accuracy as our historical dataset grows.

Limitations & Disclaimer

What we can't predict: • Weather — rain can suppress crowds 20–40%, but weather forecasts beyond 7 days are unreliable • Flash sales & ticket promotions — Disney occasionally runs unannounced deals that spike attendance • Attraction breakdowns — major ride closures redistribute crowds unpredictably • External events — nearby concerts, conventions, or sports events Park Outlook is an independent project and is not affiliated with, endorsed by, or associated with The Walt Disney Company or any of its subsidiaries. All attraction names and park references are trademarks of their respective owners. Wait times are estimates provided by third-party sources and may not reflect actual conditions. Predictions are probabilistic and should be used as general guidance only.