Search code examples
pythonpandaspermutation

Creating Permutations from DataFrame without Repetition

I've searched for a solution to this problem but haven't found anything specific to this problem. My dataframe is structured like this:

   column_1    column_2     column_3
a     2           3            7
b     9           4            3
c     1           5            2

I want to find all permutations of the above dataframe without repeating rows or columns in each individual permutation.

The preceding isn't super clear, so here is the output I'm trying to achieve:

Out: [(2,4,2),(2,5,3),(9,3,2),(9,5,7),(1,3,3),(1,4,7)]

In other words, I expected n! results

The solution I tried was:

permutations = list(product(df['column_1'], df['column_2'], df['column_3']))
print(permutations)

This returns n^n combinations.

Any help is appreciated! THANKS

Solution

  • You can use itertools.permutations on the row indices and numpy indexing:

    from itertools import permutations

idx = list(permutations(range(len(df))))

df.to_numpy()[idx, np.arange(df.shape[1])].tolist()

    output:

    [[2, 4, 2], [2, 5, 3], [9, 3, 2], [9, 5, 7], [1, 3, 3], [1, 4, 7]]