shaping

[DaveBackus]

There's a question how to drop levels of the index or columns. SO suggests

http://stackoverflow.com/questions/22233488/pandas-drop-a-level-from-a-multi-level-column-index

What we're doing here is overwriting the index or columns, which seems a little indirect, but I guess it works. In our example, we can do this with

ddi.index = ddi.index.droplevel(level='ISO')
ddt = ddi.T
ddt.columns = ddt.columns.droplevel(level='Units')

There's also a drop method that seems to drop both either rows or columns; eg,

ddi.index = ddi.index.drop('Surplus')

The column version seems to do the same thing as drop applied to a df, which seems more direct to me.

[DaveBackus]

Add sort_index to the list. Critical for speed, as the pink warning suggests.