How to use composition with pandas DataFrame and Series, not inheritance

Question

I was very tempted to inherit and subclass pandas DataFramefor my use case, but from all that I read about good programming and Pandas documentation on internal documents is the wrong thing.

As a simplified example, I can say that my lines DataFrameare people, and the fields are height and weight. I am prone to subclassing so I have a method bmi(). I would do it as follows:

import pandas as pd

class People(pd.DataFrame):
    def __init__(self, *args, **kwargs):
        super(People, self).__init__(*args, **kwargs)

    def bmi(self):
        return self['weight'] / self['height'] ** 2 x 703

people = People([[172, 74], [100, 60]])
bmi = people.bmi()

How can I do this using composition? It would be something like:

class People(object):
    def __init__(self, data):
        self.data = pd.DataFrame(data)

    def bmi(self):
        return self.data['weight'] / self.data['height'] ** 2 x 703

people = People([[172, 74], [100, 60]])
bmi = people.bmi()

If not, how do I do this? If so, why is it better?

Note. I understand that this is a simplified example.