# Split train and test from the new dataset with clustering labels from part #4
X = df_emotions_labeled.drop('sentiment', axis=1)
y = df_emotions['sentiment']

# Create oversampler (prevously defined)
# oversampler = RandomOverSampler()
# Fit and transform the data
X_resampled_labeled, y_resampled_labeled = oversampler.fit_resample(X, y)

# Split the dataset into training and testing sets with stratified sampling
X_train_lbl, X_test_lbl, y_train_lbl, y_test_lbl = train_test_split(X_resampled_labeled, y_resampled_labeled, test_size=0.2, stratify=y_resampled_labeled, random_state=42)

# Concatenate them
train_lbl = pd.concat([X_train_lbl, y_train_lbl], axis=1)
test_lbl = pd.concat([X_test_lbl, y_test_lbl], axis=1)