Issue
I am working on a small NN in keras for multi-class classification problem. I have 9 different labels and my features are also 9.
My train/test shapes are the following:
Sets shape:
x_train shape: (7079, 9)
y_train shape: (7079,)
x_test shape: (7079, 9)
y_test shape: (7079,)
But when I try to make them categorical:
y_train = tf.keras.utils.to_categorical(y_train, num_classes=9)
y_test = tf.keras.utils.to_categorical(y_test, num_classes=9)
I get the following error:
IndexError: index 9 is out of bounds for axis 1 with size 9
Here is more info about the y_train
print(np.unique(y_train)) # [1. 2. 3. 4. 5. 6. 7. 8. 9.]
print(len(np.unique(y_train))) # 9
Anyone would know what the problem is?
Solution
The shape of the y_train is 1D. You have to make it one-hot encoded. Something like
y_train = tf.keras.utils.to_categorical(y_train , num_classes=9)
And same goes for y_test too.
Update
According to the doc,
tf.keras.utils.to_categorical(y, num_classes=None, dtype="float32")
Here, y: class vector to be converted into a matrix (integers from 0 to num_classes). As in your case, y_train is something like [1,2,..]. You need to do as follows:
y_train = tf.keras.utils.to_categorical(y_train - 1, num_classes=9)
Here is an example for reference. If we do
class_vector = np.array([1, 1, 2, 3, 5, 1, 4, 2])
print(class_vector)
output_matrix = tf.keras.utils.to_categorical(class_vector,
num_classes = 5, dtype ="float32")
print(output_matrix)
[1 1 2 3 5 1 4 2]
---------------------------------------------------------------------------
IndexError Traceback (most recent call last)
<ipython-input-15-69c8be7a0f1a> in <module>()
6 print(class_vector)
7
----> 8 output_matrix = tf.keras.utils.to_categorical(class_vector, num_classes = 5, dtype ="float32")
9 print(output_matrix)
/usr/local/lib/python3.7/dist-packages/tensorflow/python/keras/utils/np_utils.py in to_categorical(y, num_classes, dtype)
76 n = y.shape[0]
77 categorical = np.zeros((n, num_classes), dtype=dtype)
---> 78 categorical[np.arange(n), y] = 1
79 output_shape = input_shape + (num_classes,)
80 categorical = np.reshape(categorical, output_shape)
IndexError: index 5 is out of bounds for axis 1 with size 5
To solve this, we convert the data to a zero-based format.
output_matrix = tf.keras.utils.to_categorical(class_vector - 1,
num_classes = 5, dtype ="float32")
print(output_matrix)
[[1. 0. 0. 0. 0.]
[1. 0. 0. 0. 0.]
[0. 1. 0. 0. 0.]
[0. 0. 1. 0. 0.]
[0. 0. 0. 0. 1.]
[1. 0. 0. 0. 0.]
[0. 0. 0. 1. 0.]
[0. 1. 0. 0. 0.]]
Answered By – M.Innat
This Answer collected from stackoverflow, is licensed under cc by-sa 2.5 , cc by-sa 3.0 and cc by-sa 4.0