How to iterate over MultiIndex levels in Pandas?

问题

I often have MultiIndex indices and I'd like to iterate over groups where higher level indices are equal. It basically looks like

from random import choice
import pandas as pd
N = 100
df = pd.DataFrame([choice([1, 2, 3]) for _ in range(N)],
                  columns=["A"],
                  index=pd.MultiIndex.from_tuples([(choice("ab"), choice("cd"), choice("de")) 
                                                   for _ in range(N)]))

for idx in zip(df.index.get_level_values(0), df.index.get_level_values(1)):
    df_select = df.ix[idx]

Is there a way to do the for loop iteration more neatly?

回答1:

Use groupby. The index of the df_select view includes the first two level values, but otherwise is similar to your example.

for idx, df_select in df.groupby(level=[0, 1]):
    ...

回答2:

Alternatively to groupby logic you can use a lambda function, which has the advantage of not having to specify the number of levels, i.e. it will pick all levels except the very last one:

for idx in df.index.map(lambda x: x[:-1]):
 df_select=df.ix[idx]

来源：https://stackoverflow.com/questions/34139121/how-to-iterate-over-multiindex-levels-in-pandas