删除基于索引的行熊猫数据框（多个条件）（Python 3.5.1）

Question

Suppose I have a Pandas DataFrame with MultiIndex on rows. How can I delete rows based on the value of one of the levels of the index based on multiple criteria?

For example, suppose I have

import pandas as pd

df = {'population': [100, 200, 300, 400, 500, 600, 700, 800]}
arrays = [['NJ', 'NJ', 'NY', 'NY', 'CA', 'CA', 'NV', 'NV'],
          ['A', 'B', None, 'D', 'E', 'F', None, 'G']]
tuples = list(zip(*arrays))
index = pd.MultiIndex.from_tuples(tuples, names=['state', 'county'])

df = pd.DataFrame(df, index=index)

                   population
state   county  
NJ        A          100
          B          200
NY        NaN        300
          D          400
CA        E          500
          F          600
NV        NaN        700
          G          800   

I want to delete all rows where the county level of the index is NaN and also delete it when it is equal to 'D' and 'G'. In other words, I want to end up with a DataFrame

                   population
state   county  
NJ        A          100
          B          200
          D          400
CA        E          500
          F          600  

So the following sort of works:

df = df.iloc[df.index.get_level_values('county') != 'D']
df = df.iloc[df.index.get_level_values('county') != 'G']

But the problem is that in my real use case there is several of these criteria. Also, I can't seem to find a way to delete NaN's using this method.

Thanks!

Answer 1

Call drop and pass a list on level='county to drop row labels with those values on that index level:

In [284]:
df.drop(['D','G',np.NaN], level='county')

Out[284]:
              population
state county            
NJ    A              100
      B              200
CA    E              500
      F              600

Answer 2

You could try using the inverse operator (~) on boolean indexing. For example,

import numpy as np
df[~(df.index.get_level_values('county').isin(['A', 'B', np.nan]))]

this line of code says "select from df where county is NOT in some list"