Grouping Data

Which city sells more? One average over every row cannot say. Split the rows into piles, one per city, then average each pile:

One line does both:

shops = pd.DataFrame({
    "city": ["Lyon", "Paris", "Lyon", "Nice", "Paris", "Lyon"],
    "sales": [20, 35, 26, 18, 41, 23],
})
means = shops.groupby("city")["sales"].mean()
print(means)

This prints:

city
Lyon     23.0
Nice     18.0
Paris    38.0
Name: sales, dtype: float64

groupby("city") makes the piles, ["sales"] picks the column, .mean() averages each pile. .sum(), .min() and .max() work too.

To rank the piles, sort them: means.sort_values(ascending=False).index[0] is Paris, the best seller.