为什么我的 .drop 删除了数据框中的所有值？

Question

我一直尝试使用这行代码来删除特定列中带有 NaN 的数据行，但它一直在删除所有行：

df = df.drop(df[(df.test_variable == 'NaN')].index)

Answer 1

To answer the question you're really trying to ask (how to drop rows containing NaNs), use the DataFrame.dropna() function:

import pandas as pd
import numpy as np

df = pd.DataFrame(np.arange(12).reshape(3,4),
    columns = ['A','B','C','D'])

df.loc[0, 'A'] = None
df.loc[1, 'B'] = None
df.loc[1, 'C'] = None

print(df)
print('------')
print(df.dropna())

prints:

$ py na.py
     A    B     C   D
0  NaN  1.0   2.0   3
1  4.0  NaN   NaN   7
2  8.0  9.0  10.0  11
------
     A    B     C   D
2  8.0  9.0  10.0  11

To answer the question you asked, though, this boolean check is problematic:

df.test_variable == 'NaN'

This will literally use the string "NaN" when checking for rows that match, instead of checking for actual NaNs. In fact, NaN is an equivalent way of saying None in Pandas, so you can use None instead of "NaN" . However, simply replacing your boolean check with df.test_variable is None will still not work, because that boolean check is literally asking if df.test_variable (which will return a Pandas Series object) is None (which it is not), instead of asking which of the elements of df.test_variable is equal to None . To do the element-wise check, use the isna() function of the Pandas Series object:

Start by getting the indices of the rows containing NaN values for a particular column (in this case, A):

df[df.A.isna()].index

Then pass the result to the drop() function:

df.drop(df[df.A.isna()].index)