The train/test split separates the data into a training set, used to fit the model, and a test set, set aside to estimate its performance on new data. You set random_state for reproducibility and use stratify in classification to preserve the class proportions.
Example
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)