从列表中生成 2×2 元组并在 python 中查找重复的元组

Question

I am a beginner in Python and I am having trouble generating and identifying duplicates on tuples on my dataFrame.

First I have this list of userid:

'userid': ["us1", "us2", "us1", "us2", "us4", "us4", "us5", "us1", "us2"]

And I want to generate 2-by-2 tuples at the order the userid are in the list, so it would be:

[('us1', 'us2'),
 ('us2', 'us1'),
 ('us1', 'us2'),
 ('us2', 'us4'),
 ('us4', 'us4'),
 ('us4', 'us5'),
 ('us5', 'us1'),
 ('us1', 'us2')]

But the tuples I arrive are this ones (and I don't understand why):

 [('us1', 'us2'),
 ('us2', 'us1'),
 ('us1', 'us4'),
 ('us4', 'us2'),
 ('us2', 'us5'),
 ('us5', 'us4'),
 ('us4', 'us1'),
 ('us1', 'us2')]

Here is my code:

   d = {'id': ["a", "a", "a", "a", "a", "a", "a", "a", "a"], 'id2': ["b", "b", "b", "b", "b", "b", "b", "b", "b"], 'userid': ["us1", "us2", "us1", "us2", "us4", "us4", "us5", "us1", "us2"], "time": [1, 2, 3, 5, 4, 7, 6, 8, 9]}
    df_test = pd.DataFrame(data=d).sort_values('time')
    df_test.groupby(['id','id2']).agg(lambda x: x.tolist()).reset_index()
    test2 = list(zip(df_test.userid[:-1], df_test.userid[1:]))
    zipped_list = test2[:]
    list(test2)

-> In addition, my next step will be finding duplicates on this tuples and extracting them for a new list, so in the case of the tuple:

    [('us1', 'us2'),
     ('us2', 'us1'),
     ('us1', 'us2'),
     ('us2', 'us4'),
     ('us4', 'us4'),
     ('us4', 'us5'),
     ('us5', 'us1'),
     ('us1', 'us2')]

Should be the list [('us1', 'us2'), 3] because is the only tuple that appears duplicated and the '3' is to say that appears 3 times this duplication.

Therefore I cannot find my error on generating the tuples on the order I want nor having any idea on how to find the duplicates.

Answer 1

Let us do frozenset + value_counts

pd.Series(list(map(frozenset,zipped_list))).value_counts()
(us2, us1)    3
(us1, us4)    2
(us2, us5)    1
(us5, us4)    1
(us2, us4)    1
dtype: int64

If only need the list reorder

l=list(map(frozenset,zipped_list))

Or we can do numpy

np.sort(zipped_list,axis=1).tolist()
[['us1', 'us2'], ['us1', 'us2'], ['us1', 'us4'], ['us2', 'us4'], ['us2', 'us5'], ['us4', 'us5'], ['us1', 'us4'], ['us1', 'us2']]

Update: you sort_values first, so we need sort_index back

list(zip(df_test.userid[:-1].sort_index(), df_test.userid[1:].sort_index()))
[('us1', 'us2'), ('us2', 'us1'), ('us1', 'us2'), ('us2', 'us4'), ('us4', 'us4'), ('us4', 'us5'), ('us5', 'us1'), ('us1', 'us2')]