Can I use xor on numpy matrices?

Question

I have 2 numpy matrix like this.

matrix1

arr1 =
array([[ 0.,  0.,  0.],
       [ 0.,  0.,  0.],
       [ 0.,  1.,  0.]])

matrix2

arr2 =
array([[ 0.,  0.,  0.],
       [ 0.,  0.,  1.],
       [ 0.,  0.,  0.]])

I want to find similarity of these matrices. I think xor can be used on matrices. Xor operation should show where values are different and then I can count value 1 to calculate a percentage of similarity. I don't know how to use xor in python.

This code doesn't work: a = arr1 xor arr2 .

Answer 1

You can simply use arr1 != arr2 which results in:

>>> arr1 != arr2
array([[False, False, False],
       [False, False,  True],
       [False,  True, False]], dtype=bool)

and then use .sum() since int(False) is 0 and int(True) is 1:

>>> (arr1 != arr2).sum()
2

So there are two indices for which arr1[i,j] is not equal to arr2[i,j].

If you want to calculate the similarity (here defined as the number of elements that are the same) you can use:

>>> (arr1 == arr2).sum()/arr1.size
0.77777777777777779

so 77.77% of the elements are the same.

Answer 2

There is also the inbuilt function bitwise_xor