训练和验证数据集的拆分

Question

I need to split my training data (80-20) into validation data in a way that the split sub-datasets are not random but always the same.

Presently I use this code

from sklearn.model_selection import train_test_split
X_train, X_val, Y_train, Y_val = train_test_split(X, Y, test_size=0.2)

but the split sub-datasets are always random and never the same. I want it to be random but the same value should be present when I run the code again ( something like np.random.seed)

Is there a way to do that?

Answer 1

train_test_split() has a random_state argument. If you assign to it an integer value the result will be always the same:

from sklearn.model_selection import train_test_split
X_train, X_val, Y_train, Y_val = train_test_split(X, Y, test_size=0.2, random_state=1)