course content

Course Content

Data Manipulation using pandas

Grouping by Several ColumnsGrouping by Several Columns

Is it possible to group by pairs of values? For instance, we can group by countries and then by their regions. Yes, it's also possible in pandas! To group by several columns, use the same .groupby() method passing list of columns that will be used to determine groups. How does such a grouping work? Look at the picture below.

Picture

As you can see, at first values were grouped by 'Group' and then by 'Subgroup' among each of groups. For instance, let's find out number of households for each pair of 'roomh', 'hhsize' columns values (number of rooms and number of people in a dwelling, respectively).

The output is quite big, since number of possible combinations is quite large. For instance, you can see that there are 59 dwellings with 10 or more rooms with 4 people living in it.

Everything was clear?

Section 3. Chapter 6