如何创建具有不丢失数据（不是 NaN）的特征、值和索引的 DataFrame？

Question

I have the following Data Frame

data = {'first_set_of_numbers':  [3,9,6,np.nan],
        'second_set_of_numbers': [np.nan,13,np.nan,np.nan]
       }
df = pd.DataFrame(data,columns=['first_set_of_numbers','second_set_of_numbers'], index=['A','B','C','D'])
df

How can I now get a new Data Frame that shows all the not missing values with their accoring feature and index? It should look like this:

Answer 1

You can use df.stack() and df.reset_index :

In [2493]: df.stack().reset_index(level=[1], name='value').rename(columns={'level_1':'feature'})
Out[2493]: 
                 feature  value
A   first_set_of_numbers    3.0
B   first_set_of_numbers    9.0
B  second_set_of_numbers   13.0
C   first_set_of_numbers    6.0

Answer 2

You can use df.melt with ignore_index parameter set to False and use df.dropna here.

df.(ignore_index=, var_name='features', value_name='value').() 
# default values of `var_name` -> 'variable', `value_name`->'value'

                features  value
A   first_set_of_numbers    3.0
B   first_set_of_numbers    9.0
C   first_set_of_numbers    6.0
B  second_set_of_numbers   13.0