Grouping and summarising

Datasets usually arrive long: one row per event, many rows per thing you care about. Six temperature readings, three cities. Group by turns that into one row per city.

readings.groupby("city")["temperature"].mean()

Read it left to right:

1. groupby("city") — put the rows into piles, one pile per distinct city
2. ["temperature"] — from each pile, take that column
3. .mean() — reduce each pile to a single number

What comes back is a Series indexed by city:

city
Cairo     35.000000
Lisbon    22.666667
Oslo       5.500000
Name: temperature, dtype: float64

Because the index is the city name, you look values up by name — by_city["Lisbon"] — and .idxmax() gives you the label of the largest value rather than its position.

.mean() is only one option: .sum(), .max(), .count() and .min() all work the same way.

1. Create readings — a DataFrame with the columns city and temperature
2. Create by_city — the average temperature per city, using groupby
3. Create warmest — the name of the warmest city, as a str
4. Create lisbon_average — Lisbon's average, as a float
5. Print exactly:

Warmest: Cairo
Lisbon averages 22.67