Convert array of indices to 1-hot encoded numpy array

Question

Your array a defines the columns of the nonzero elements in the output array. You need to also define the rows and then use fancy indexing:

>>> a = np.array([1, 0, 3])
>>> b = np.zeros((a.size, a.max()+1))
>>> b[np.arange(a.size),a] = 1
>>> b
array([[ 0.,  1.,  0.,  0.],
       [ 1.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  1.]])

>>> values = [1, 0, 3]
>>> n_values = np.max(values) + 1
>>> np.eye(n_values)[values]
array([[ 0.,  1.,  0.,  0.],
       [ 1.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  1.]])

In case you are using keras, there is a built in utility for that:

from keras.utils.np_utils import to_categorical   

categorical_labels = to_categorical(int_labels, num_classes=3)

And it does pretty much the same as @YXD's answer (see source-code).

Here is what I find useful:

def one_hot(a, num_classes):
  return np.squeeze(np.eye(num_classes)[a.reshape(-1)])

Here num_classes stands for number of classes you have. So if you have a vector with shape of (10000,) this function transforms it to (10000,C). Note that a is zero-indexed, i.e. one_hot(np.array([0, 1]), 2) will give [[1, 0], [0, 1]].

Exactly what you wanted to have I believe.

PS: the source is Sequence models - deeplearning.ai

You can also use eye function of numpy:

numpy.eye(number of classes)[vector containing the labels]

You can use sklearn.preprocessing.LabelBinarizer:

Example:

import sklearn.preprocessing
a = [1,0,3]
label_binarizer = sklearn.preprocessing.LabelBinarizer()
label_binarizer.fit(range(max(a)+1))
b = label_binarizer.transform(a)
print('{0}'.format(b))

output:

[[0 1 0 0]
 [1 0 0 0]
 [0 0 0 1]]

Amongst other things, you may initialize sklearn.preprocessing.LabelBinarizer() so that the output of transform is sparse.