Train test split before or train test split inside cross validation

Question

I want to do cross validation. So should i split my data into train and test with sklearn train_test_split and use cross validation like this:

cross_validate(model, X_train, y_train, scoring = roc_auc, cv = 5, n_jobs = -1)

Or should i make a function and split the data inside each fold differently like this:

def cross_val_evaluation(model):
    kf = KFold(n_splits=5)
    for train_index, test_index in kf.split(X,y):
        X_train, X_test = X.iloc[train_index, :], X.iloc[test_index, :]
        y_train, y_test = y.iloc[train_index], y.iloc[test_index]  
        y_pred_proba = model.predict_proba(X_test)
        cross_val_roc_auc_score = round(roc_auc_score(y_test,y_pred_proba[:,1]),3)
        cross_roc_auc_score_lst_test.append(cross_val_roc_auc_score_test)
    print(f'AUC Score Cross_Val Proba Test: {round(np.mean(cross_roc_auc_score_lst_test),3)}') 

Answer 1

First, split the dataset into train and test dataset. Then, cross-validate on the train dataset.

You should use the scikit-learn functions. Implementing the functionality yourself may introduce bugs.