如何通过总结特定列来“压扁”熊猫面板

Question

I am still familairizing myself with Pandas, and Python in general, so please excuse if this is a simple question. I'd also like to avoid one liners so I can understand the underlying actions if possible! :)

I've managed to pull data data which results in a Panel, with four items. The key of each item is a calendar quarter:

Item '2015-03-31':
Type        Quarterly Sales        Ending Inventory
Shoes       123,456                50,000
Purses      33,222                 10,000

Item '2015-06-30':
Type        Quarterly Sales        Ending Inventory
Shoes       12,744                 56,000
Purses      15,123                 9,000

Item '2015-9-30':
Type        Quarterly Sales        Ending Inventory
Shoes       15,998                 35,000
Purses      11,222                 15,000

Item '2015-12-31':
Type        Quarterly Sales        Ending Inventory
Shoes       12,000                 45,000
Purses      9,551                  7,000

Ultimately, I would like to "flatten" this by summing up the Quarterly Sales , but taking the Type and Ending Inventory from the most recent entry, and have this in a DataFrame. So my ending DataFrame would be something like this:

Type        Quarterly Sales        Ending Inventory
Shoes       164,198                45,000
Purses      69,118                 7,000

I tried using a function such as grouby (eg mypanel.groupby('Type').sum() ), but that ended up summing both Quarterly Sales and Ending Inventory , whereas I want to take the "most recent" Ending Inventory instead. An easy "fix" for this would be to take the resulting DataFrame, and then subtract out the summation of the first three quarters for the Ending Inventory column, but that seems incredibly awkward.

Any suggestions?

Thanks!

Answer 1

agg_dict = {'Quarterly Sales': 'sum', 'Ending Inventory': 'last'}
pnl.to_frame().T.stack(0).groupby(level='Type').agg(agg_dict)