df.isnull().sum() 
      //type: <class 'pandas.core.series.Series'>

or

df.column_name.isnull().sum()
     //type: <type 'numpy.int64'>

根据投票最多的答案，我们可以很容易地定义一个函数，给我们一个数据框架来预览每一列中缺失的值和缺失值的百分比:

def missing_values_table(df):
    mis_val = df.isnull().sum()
    mis_val_percent = 100 * df.isnull().sum() / len(df)
    mis_val_table = pd.concat([mis_val, mis_val_percent], axis=1)
    mis_val_table_ren_columns = mis_val_table.rename(
    columns = {0 : 'Missing Values', 1 : '% of Total Values'})
    mis_val_table_ren_columns = mis_val_table_ren_columns[
        mis_val_table_ren_columns.iloc[:,1] != 0].sort_values(
    '% of Total Values', ascending=False).round(1)
    print ("Your selected dataframe has " + str(df.shape[1]) + " columns.\n"      
        "There are " + str(mis_val_table_ren_columns.shape[0]) +
            " columns that have missing values.")
    return mis_val_table_ren_columns

你可以试试:

In [1]: s = pd.DataFrame('a'=[1,2,5, np.nan, np.nan,3],'b'=[1,3, np.nan, np.nan,3,np.nan])

In [4]: s.isna().sum()   
Out[4]: out = {'a'=2, 'b'=3} # the number of NaN values for each column

如果需要nan的总和:

In [5]: s.isna().sum().sum()
Out[6]: out = 5  #the inline sum of Out[4] 

自从pandas 0.14.1以来，我的建议在value_counts方法中有一个关键字参数已经实现:

import pandas as pd
df = pd.DataFrame({'a':[1,2,np.nan], 'b':[np.nan,1,np.nan]})
for col in df:
    print df[col].value_counts(dropna=False)

2     1
 1     1
NaN    1
dtype: int64
NaN    2
 1     1
dtype: int64

df.isnull().sum() 
      //type: <class 'pandas.core.series.Series'>

or

df.column_name.isnull().sum()
     //type: <type 'numpy.int64'>

另一种完整的方法是使用np。带有.isna()的count_non0:

np.count_nonzero(df.isna())

%timeit np.count_nonzero(df.isna())
512 ms ± 24.7 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

使用1000005行× 16列的数据框架与顶部答案进行比较:

%timeit df.isna().sum()
492 ms ± 55.2 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

%timeit df.isnull().sum(axis = 0)
478 ms ± 34.9 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

%timeit count_nan = len(df) - df.count()
484 ms ± 47.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

数据:

raw_data = {'first_name': ['Jason', np.nan, 'Tina', 'Jake', 'Amy'], 
        'last_name': ['Miller', np.nan, np.nan, 'Milner', 'Cooze'], 
        'age': [22, np.nan, 23, 24, 25], 
        'sex': ['m', np.nan, 'f', 'm', 'f'], 
        'Test1_Score': [4, np.nan, 0, 0, 0],
        'Test2_Score': [25, np.nan, np.nan, 0, 0]}
results = pd.DataFrame(raw_data, columns = ['first_name', 'last_name', 'age', 'sex', 'Test1_Score', 'Test2_Score'])

# big dataframe for %timeit 
big_df = pd.DataFrame(np.random.randint(0, 100, size=(1000000, 10)), columns=list('ABCDEFGHIJ'))
df = pd.concat([big_df,results]) # 1000005 rows × 16 columns