I'm trying

print(Y)
print(Y.shape)class_weights = compute_class_weight('balanced',np.unique(Y),Y)
print(class_weights)

But this gives me an error:

ValueError: classes should include all valid labels that can be in y

My Y looks like:

       0  1  2  3  4
0      0  0  1  0  0
1      1  0  0  0  0
2      0  0  0  1  0
3      0  0  1  0  0
...
14992     0  0  1  0  0
14993      0  0  1  0  0

And my Y.shape looks like: (14993, 5)

In my keras model, I want to use the class_weights as it is an uneven distribution:

model.fit(X, Y, epochs=100, shuffle=True, batch_size=1500, class_weights=class_weights, validation_split=0.05, verbose=1, callbacks=[csvLogger])

Just transform the one-hot encoding to categorical labels:

from sklearn.utils import class_weighty = Y.idxmax(axis=1)class_weights = class_weight.compute_class_weight('balanced',np.unique(y),y)# Convert class_weights to a dictionary to pass it to class_weight in model.fit
class_weights = dict(enumerate(class_weights))