whenever i work with pandas i try to use an ipython style console so that I can mess about with the dataframe quickly and then later solidify what I did into code.
that, and a combination of stackexchange and pandas documentation
I rated this free online book which has a chapter on numpy and a chapter on pandas:
https://jakevdp.github.io/PythonDataScienceHandbook/
the set up for groupby is usually the same.
data = sorted(data, key=sorter)
grouped = groupby(data, key=sorter)
groupby is a generator object
iterating through groupby gives you (key, group) pairs.
a group is also a generator object
iterating through the group gives you the data with that key
that’s why in all the code with itertools groupby you have to iterate through the group generator object.