tf.keras.utils.to_categorical

TensorFlow 1 version View source on GitHub

Converts a class vector (integers) to binary class matrix.

E.g. for use with categorical_crossentropy.

Usage Example:

y = [0, 1, 2, 3]
tf.keras.utils.to_categorical(y, num_classes=4)
array([[1., 0., 0., 0.],
       [0., 1., 0., 0.],
       [0., 0., 1., 0.],
       [0., 0., 0., 1.]], dtype=float32)

y class vector to be converted into a matrix (integers from 0 to num_classes).
num_classes total number of classes.
dtype The data type expected by the input. Default: 'float32'.

A binary matrix representation of the input. The classes axis is placed last.