The same five steps for every day of every destination. No AI writes a score — pure, published math over sourced signals.
Cruise ships from the port's published schedule, each matched to its real size (guest capacity at double occupancy — we count berths, never actual passenger headcounts). Public + school holidays for every visitor country and always the destination's own. Seasonality from official overnight statistics where available.
Cruise 45% · season 35% · holidays 20%. These are our published choices — never claimed as measured. A confirmed event can only ever raise a day (a floor), and every floor is shown on the day itself.
A day's raw pressure is ranked against the same destination's other 364 days — never against another city, never a world average. A quiet island's "Busy" stays honest.
Calm (bottom 40%) · Lively (40–65%) · Busy (65–85%) · Packed (top 15%).
● measured real, sourced value · ◐ computed deterministic math on measured inputs · ⚑ estimated a labeled stand-in, never counted as measured · ○ no data a hole is a hole — never filled with a guess. A day is scored from whichever signals are present, reweighted over what's there; a day with no usable signal at all publishes no forecast. Ship-size coverage below 70% turns cruise into "no data" rather than a guessed magnitude. A measured zero is not a gap: "No ships scheduled — confirmed by the port schedule" is a real value from the schedule; "○ no data" publishes nothing. Quiet weeks with harsh weather are labeled quiet but harsh and never badged Recommended. Ferry day-trippers aren't counted yet: treat every band as a floor.
State FRESH · ship sizes 89.5% matched · artifact b50005e88c9358ed · model v10-2026.07 · computed 22 Jul 2026 13:47. Every published forecast is frozen in an append-only ledger — what we predicted, as of when, can be checked later and never rewritten.
Monthly typical_temp_max_c / typical_rain_days are climate normals from a meteorological archive (2019–2023 daily means) — climate, never a forecast. This destination's normals: meteorological archive:41.39,2.17:2019-2023.
Questions the page can't answer are honest gaps, not omissions. Data license CC BY 4.0 — cite Avoid-Crowds with a link.
Credits: Weather data by Open-Meteo.com (CC BY 4.0, aggregated into monthly normals) · school-holiday windows contain information from the OpenHolidays API, made available under ODbL 1.0.