Python list into numpy array

Question

I have seen lots of example of lists into arrays, but no examples of lists in this format into arrays, which is strange because the list format I present is the standard go-to way of defining a graph, point-to-point mapping that you would find in any table, csv, database, etc. I tried everything here with no luck. Thanks for any ideas.

input=  [[A, A, 0],
        [A, B, 5],
        [A, C, 3],
        [B, A, 5],
        [B, B, 0],
        [B, C, 6],
        [C, A, 3],
        [C, B, 6],
        [C, C, 0]]

desiredOutput=  [[0, 5, 3],
                [5, 0, 6],
                [3, 6, 0]]

input= [[A, A, 0], [A, B, 5], [A, C, 3], [B, A, 5], [B, B, 0], [B, C, 6], [C, A, 3], [C, B, 6], [C, C, 0]] desiredOutput= [[0, 5, 3], [5, 0, 6], [3, 6, 0]] — user3431083
– user3431083, Commented Oct 5, 2017 at 14:59
I think a module like pandas is often the way you can handle such data. — Julien Spronck
– Julien Spronck, Commented Oct 5, 2017 at 15:15

PM 2Ring · Accepted Answer · 2017-10-05 15:22:44Z

2

Here's one way to produce your adjacency matrix as a 2D Numpy array. It assumes that the input graph data is correct, in particular, that its length is a perfect square.

import numpy as np

graph_data =  [
    ['A', 'A', 0], ['A', 'B', 5], ['A', 'C', 3],
    ['B', 'A', 5], ['B', 'B', 0], ['B', 'C', 6],
    ['C', 'A', 3], ['C', 'B', 6], ['C', 'C', 0],
]

size = np.sqrt(len(graph_data)).astype(np.int)
adjacency_matrix = np.array(graph_data)[:,-1].astype(np.int).reshape(size, size)
print(adjacency_matrix)

output

[[0 5 3]
 [5 0 6]
 [3 6 0]]

The above code also assumes that the graph data is in the proper order, since it ignores the letters. Of course, that's easily handled by sorting the graph data before attempting to convert it to an array. Eg,

graph_data.sort()

Here's a pure Python version that outputs a list of tuples:

graph_data =  [
    ['A', 'A', 0], ['A', 'B', 5], ['A', 'C', 3],
    ['B', 'A', 5], ['B', 'B', 0], ['B', 'C', 6],
    ['C', 'A', 3], ['C', 'B', 6], ['C', 'C', 0],
]

graph_data.sort()
size = int(len(graph_data) ** 0.5)
it = iter(row[-1] for row in graph_data)
print(list(zip(*[it]*size)))

output

[(0, 5, 3), (5, 0, 6), (3, 6, 0)]

edited Oct 5, 2017 at 15:22

answered Oct 5, 2017 at 15:17

PM 2Ring

55.6k6 gold badges96 silver badges201 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

user3431083 Over a year ago

This solution was selected because it was exponentially faster than the other. Specifically, the solution which outputs a list of tuples, which then can be converted into a numpy array. Tested performance with a 1.8 million node example - the other suggestion took 10 seconds, but this solution was an amazing 0.3 seconds

Julien Spronck · Accepted Answer · 2017-10-05 15:08:04Z

2

You can just slice your array and reshape it:

input= [['A', 'A', 0], ['A', 'B', 5], ['A', 'C', 3], ['B', 'A', 5], ['B', 'B', 0], ['B', 'C', 6], ['C', 'A', 3], ['C', 'B', 6], ['C', 'C', 0]]
arr = np.array(input)
desiredOutput=arr[:, 2].reshape(3, 3).astype(np.float)
# array([[ 0.,  5.,  3.],
#        [ 5.,  0.,  6.],
#        [ 3.,  6.,  0.]])

edited Oct 5, 2017 at 15:08

answered Oct 5, 2017 at 15:06

Julien Spronck

15.5k5 gold badges50 silver badges57 bronze badges

3 Comments

tobias_k Over a year ago

Provided that the entries in input are always in the correct order, i.e. the first two columns do not actually matter... (but if that's not the case, you could just sort the input)

user3431083 Over a year ago

Order is always alphabetical: order by col 1, then col 2.

Julien Spronck Over a year ago

@tobias_k you're right ... it might not be the case.

Collectives™ on Stack Overflow

Python list into numpy array

2 Answers 2

1 Comment

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related