Pandas - flattening a multiindex column containing tuples, but ignore missing values

Question

I have a multiindex pandas dataframe like this:

lst = [(1, 2), (3, 4), (5, 6), (7, 8), (9, 10), (11, 12), (13, 14), (21, 22)]
df = pd.DataFrame(lst, pd.MultiIndex.from_product([['A', 'B'], ['1','2', '3', '4']])).loc[:('B', '2')]
df["tuple"] = list(zip(df[0], df[1]))

#df:
      0   1     tuple
A 1   1   2    (1, 2)
  2   3   4    (3, 4)
  3   5   6    (5, 6)
  4   7   8    (7, 8)
B 1   9  10   (9, 10)
  2  11  12  (11, 12)

I want to transform the column, containing the tuples, into a list of tuples. My approach is:

#dataframe to append list of tuples
new_df = pd.DataFrame([1, 2], index = list("AB") )

#voila a list of tuples
new_df["list_of_tuples"] = df["tuple"].unstack(level = -1).values.tolist()

#new_df:
   0                 list_of_tuples
A  1     [(1, 2), (3, 4), (5, 6), (7, 8)]
B  2  [(9, 10), (11, 12), None, None]

This works, but only for multiindex dataframes with equal length for each entry. If all entries don't have the same length, the missing columns give rise to a None value in the list. My attempts to remove numpy NaN values, before creating a list, failed. Is there an approach to prevent the appearance of None in the final list of tuples?

Answer 1

Is this what you need ?

df.groupby(level=[0]).tuple.apply(list)
Out[306]: 
A    [(1, 2), (3, 4), (5, 6), (7, 8)]
B                 [(9, 10), (11, 12)]
Name: tuple, dtype: object