About Kairoto

Kairoto answers one question well: when should you visit a particular place? Most of what's online is generic - a 2018 listicle that treats "best time to visit" as a single fixed answer for everyone. It isn't. The right week depends on the weather you'll tolerate, how much you mind crowds, what you're willing to pay, what's happening locally, and how comfortable the conditions are. Kairoto weighs all of that across every week of the year.

The name comes from the Greek kairos - the opportune moment to act.

How it works, in short

For each destination, Kairoto precomputes a 0–100 score per factor (weather, crowds, affordability, events, comfort) for all 52 weeks, from climate normals and curated data. When you adjust the importance sliders, it combines those factor scores into a single weekly recommendation right in your browser - so the answer updates instantly - and shows you the reasoning behind it. Every recommendation is deterministic: the same inputs always give the same answer. Here's the full chain, from raw data to the number you see.

No black box, no machine learning - just five measurable factors, weighted by what you tell us matters.

1. Five factors, scored 0–100

For each of the 52 weeks of the year, every destination carries a precomputed score for five factors:

• Weather - a temperature "comfort curve" (best around 18–24 °C, decaying either side), plus a precipitation penalty (the harsher of rainfall amount and rainy-day probability), plus a daylight term. Sourced from climate normals.
• Crowds - a hand-tuned baseline per week (100 = empty, 0 = packed), with penalties layered on for public holidays, school breaks in major source markets, and big local events.
• Affordability - a curated curve reflecting flight- and accommodation-price seasonality (100 = cheapest, 0 = peak).
• Events & culture - a neutral baseline of 50, nudged up by festivals and cultural happenings and down by strikes, closures and the like.
• Comfort - starts at 100 and is docked only for real stressors: mugginess (from dew point, so dry heat isn't penalised and cool-damp air stays comfortable), air quality (US AQI), and, in dengue/malaria regions, a climate-driven disease-season adjustment. A genuinely pleasant week reads 100.

2. Your priorities become weights

Each slider sets how much one factor matters to you, from "doesn't matter" to "critical" (mapped to a weight between 0.01 and 1). Nothing ever drops to exactly zero - a factor can be made nearly irrelevant, not silenced.

3. A weighted geometric mean per week

For each week, the five factor scores are combined into one overall score using a weighted geometric mean - the factor scores raised to their slider weights, multiplied together, then taken to the power of one over the total weight. A geometric mean is dragged toward its lowest term, so a single bad factor genuinely hurts a week's score - which matches how trips actually feel.

4. Smoothing

The 52 weekly scores are passed through a centred 5-week moving average (the year wraps around, since it's cyclic). This keeps the year's shape legible without overreacting to a single noisy week. Event markers are kept separate, so a one-week festival still shows up rather than being averaged away.

5. Reasoning

The headline, the per-week detail and the "why not this week" notes are generated from templates filled with the underlying data, with phrasing picked deterministically (seeded by destination and week) so it's stable and varies naturally between places.

A few caveats

• This is climatology, not a forecast - long-run averages, not what the weather will actually do on your dates.
• Crowd and affordability curves encode judgement, informed by data but not infallible.
• Events are curated by hand and won't be exhaustive.
• The weights in the formulas (and the comfort curve) are tunable defaults, not laws of nature.
• Always verify dates, prices and conditions with official sources before booking.

Who made this

Kairoto is built by me, Manav Dodia. It's a work in progress - destinations and features are being added over time.