
        
        
            
                
                
                    df = pd.DataFrame({"A":[3,pd.np.NaN,5],"B":[6,7,pd.np.NaN]}, index=["a","b","c"])
df
                
            
               A    B
a  3.0  6.0
b  NaN  7.0
c  5.0  NaN

Solution

To get the index of rows with missing values in Pandas optimally:


        
        
            
                
                
                    temp = df.isna().any(axis=1)
temp[temp].index
                
            
            Index(['b', 'c'], dtype='object')

Explanation

We first check for the presence of NaNs using isna(), which returns a DataFrame of booleans where True indicates the presence of a NaN:


        
        
            
                
                
                    df.isna()
                
            
               A      B
a  False  False
b  True   False
c  False  True

Next, we use any(axis=1), which scans each row and returns a Series of booleans where True indicates a row with at least one True:


        
        
            
                
                
                    df.isna().any(axis=1)
                
            
            a    False
b     True
c     True
dtype: bool

We then temporarily store this intermediate result in a variable called temp. Our goal now is to extract the Index where the corresponding value is True (b and c in this case).

We first exclude indexes where the corresponding values is False by treating temp as a boolean mask:


        
        
            
                
                
                    temp[temp]
                
            
            b    True
c    True
dtype: bool

Finally, all we need to do is to access the index property of this Series:


        
        
            
                
                
                    temp[temp].index
                
            
            Index(['b', 'c'], dtype='object')

Getting integer index

Again, consider the same df as above:


        
        
            
                
                
                    df = pd.DataFrame({"A":[3,pd.np.NaN,5],"B":[6,7,pd.np.NaN]}, index=["a","b","c"])
df
                
            
               A    B
a  3.0  6.0
b  NaN  7.0
c  5.0  NaN

Solution

To get the integer indexes of rows with missing values:


        
        
            
                
                
                    np.where(df.isna().any(axis=1))[0]   # returns a NumPy array
                
            
            array([1, 2])

Explanation

Similar to the case above, we start by checking for the presence of NaN values using isna():


        
        
            
                
                
                    df.isna()   # returns a DataFrame
                
            
               A      B
a  False  False
b  True   False
c  False  True

We then check for rows where there is at least one True:


        
        
            
                
                
                    df.isna().any(axis=1)   # returns a Series
                
            
            a    False
b     True
c     True
dtype: bool

To get the integer index of the boolean True, use np.where(~):


        
        
            
                
                
                    np.where(df.isna().any(axis=1))   # returns a tuple of size one
                
            
            (array([1, 2]),)

Here, np.where(~) returns a tuple of size one, and so we use [0] to extract the NumPy array of indexes:


        
        
            
                
                
                    np.where(df.isna().any(axis=1))[0]   # returns a NumPy array
                
            
            array([1, 2])