Chapter 1 · From Data to Decisions

From data to decision: how much evidence do you need?

A decision without data is a guess. Start with nothing, then add one signal at a time — season, cloud cover, humidity — for the same 48 historical days, and watch the prediction get less arbitrary at each step. No formulas here; Chapter 3's Bayes' Theorem demo shows exactly how these signals combine mathematically.

48 past days

Every row is a real day in the record. The check/cross column shows whether this stage's rule would have gotten that day right.

48 days · 17 rainy

This is a preview, not the mechanism. Real decision rules aren't picked by eyeballing a table — Chapter 3 (Will It Rain? Bayes' Theorem) shows how prior belief and each piece of evidence combine mathematically into a posterior, and Chapter 4 (Parzen Windows & k-NN) shows the same "how much evidence is enough" question with a tunable window instead of fixed evidence columns.