can you please explain the output of this code? and how to use where function in numpy?-CodePudding

a = np.array([[1,2],[3,4]])
np.where(a<4)

answer:

array([0,0,1]), array([0,1,0])

please explain output:

answer:

array([0,0,1]),array([0,1,0])

CodePudding user response：

numpy.where gives you the indices of the truthy values.

I hope this break down helps you to understand the logic:

a = np. array([[1,2],[3,4]])
#         0  1
# array([[1, 2],   # 0
#        [3, 4]])  # 1

a<4
#            0      1
# array([[ True,  True],    0
#        [ True, False]])   1


# flat version
# row:          0     0       1      1
# col:          0     1       0      1
# # array([[ True,  True], [ True, False]])

# keep only the True
# row:  [0, 0, 1]
# col:  [0, 1, 0]

np.where(a<4)
# (array([0, 0, 1]), array([0, 1, 0]))

CodePudding user response：

np.where returns a tuple with (in this case) 2 elements.

The first element are row indices of elements meeting the condition.

The second element are column indices of elements meeting the condition.

To check it, save the result, e.g.:

ind = np.where(a < 4)

Now, when you run a[ind], you will get a (1-D) array filled with elements meeting this condition, i.e.:

array([1, 2, 3])

If your source array has more dimensions, the resulting tuple will have more components.