Should we apply feature scaling to output training set (Y_train)

Question

Should we apply feature scaling to output data in machine learning? Can we select to scale certain features in input parameters and not all of them?

from sklearn.preprocessing import StandardScaler