[Python] How to one-hot encode a vector with arbitrary number of unique values?

Given a output vector y that can have any number of discrete values, say from the set {1, 2, 3, 4}. Let the vector have this instance:

y = [1, 1, 2, 4, 2, 3, 1]


Is there a numpy-tonic library way of generating a one-hot encoded representation of this vector? I.e. such that

y_enc =
y1 y2 y3 y4
1 0 0 0
1 0 0 0
0 1 0 0
0 0 0 1
0 1 0 0
0 0 1 0
1 0 0 0


For the case {0, 1} i have a small example but I cant see this going in the right direction:

>>> k
array([[0.],
[0.],
[0.],
[0.],
[1.]])
>>> z = np.zeros((5,2))
>>> z
array([[0., 0.],
[0., 0.],
[0., 0.],
[0., 0.],
[0., 0.]])
>>> for i,ki in enumerate(k):
... if (ki == 0):
... z[0] += 1
... if (ki == 1):
... z[1] += 1
...
>>> z
array([[1., 0.],
[1., 0.],
[1., 0.],
[1., 0.],
[0., 1.]])
>>>

Continue reading...