Most examined · learn first
Data conventions
n/a data not available
tr trace = rainfall < 0.05mm — treat as 0
— a dash for visibility = missing data
The LDS has gaps → this is the answer to "why is the sample fewer than 184 / fewer than expected".
11 variables · name + unit + type
continuous discrete qualitative
Beaufort scale (windspeed)
Windspeed is also given on the Beaufort scale — know the categories (used in questions like "Moderate days"):
0 — Calm — <1 kn
1–3 — Light — 1–10 kn
4 — Moderate — 11–16 kn
5 — Fresh — 17–21 kn
Suitable for a normal distribution?
Sometimes asked: "why is X not suitable to model with a normal distribution?"
Not suitable: Cloud cover (discrete, oktas) · Rainfall (skewed, not bell-shaped) · Pressure (skewed) · Windspeed (not symmetric). Wind direction is qualitative, so it can't be modelled this way either.
8 locations · characteristics
UK 5 UK stations · OVERSEAS 3 overseas (only 4 variables)
UK stations
North → cooler. Leuchars is the most northerly = coldest; Camborne (SW, by the sea) = wettest.
Overseas stations
Only 4 variables each (temperature, rainfall, windspeed, pressure).
Perth sits below the Equator → Southern Hemisphere → seasons reversed (May–Oct = autumn→winter, so it's cold).
Context-question templates · just apply them
Red text must appear or you lose the mark.