This lesson's interactive features are locked, please to keep using them

Grouping in Pandas

groupby() is the built-in way to aggregate in Pandas. It uses the split-apply-combine pattern.

orders = pd.DataFrame(
    {
        "region": ["West", "East", "West", "East", "West"],
        "revenue": [100, 200, 150, 300, 250],
    }
)

revenue_by_region = orders.groupby("region")["revenue"].sum()

The result:

region
East    500
West    500

Here's what happened:

Split – Pandas grouped the rows by region (West rows together, East rows together)
Apply – .sum() ran on each group's revenue column
Combine – Pandas returned the results as a single Series, with the group keys (region) as its index

You can run other aggregation functions as well, once you've grouped the data:

orders.groupby("region")["revenue"].mean()  # average
orders.groupby("region")["revenue"].count()  # number of non-null values
orders.groupby("region")["revenue"].min()  # smallest value
orders.groupby("region")["revenue"].max()  # largest value

Assignment

SnackStack wants a dashboard tile showing how hot each region's devices are running on average.

Complete the average_temperature_by_region function. It accepts a DataFrame with a temperature_celsius column, and returns a tidy DataFrame with one row per region.

Group the devices by region and take the mean of the temperature_celsius column.
Round the averages to 1 decimal place.
Reset the index so region becomes a regular column.
Rename the temperature_celsius column to avg_temperature.
Sort the rows by avg_temperature and return the result.