Could anyone explain why do these two snippets of code produce different results? (Cross Validation)

Question

I am currently undergoing a machine learning project to predict default. Decided to apply logistic regression and managed to get pretty decent results, but I decided to apply cross-validation and then use parameter tunning to see if I can find better results. I wrote these two blocks of code:

accuracytest = []

recalltrain = []
recalltest = []

precisiontrain = []
precisiontest = []


f1scoretrain = []
f1scoretest = []


KF = StratifiedKFold(n_splits = 5, random_state= 42, shuffle = True)
KF.get_n_splits(X, y)

for train_index, test_index in KF.split(X, y):
    print("Train:", train_index, "Validation:", test_index)
    X_train_scaled, X_test_scaled = X.iloc[train_index], X.iloc[test_index] 
    y_train, y_test = y.iloc[train_index], y.iloc[test_index] 
    
    sgd_LogisticRegression = SGDClassifier(loss='log', n_jobs = -1, warm_start = True, class_weight = "balanced").fit(X_train_scaled, y_train)    
    
    y_pred_train = sgd_LogisticRegression.predict(X_train_scaled) 

    y_pred_test = sgd_LogisticRegression.predict(X_test_scaled) 
    
    accuracy_train = accuracy_score(y_train, y_pred_train)
    accuracy_test = accuracy_score(y_test, y_pred_test)
    
    recall_train = recall_score(y_train, y_pred_train)
    recall_test = recall_score(y_test, y_pred_test)
    
    precision_train = precision_score(y_train, y_pred_train)
    precision_test = precision_score(y_test, y_pred_test)
    
    f1score_train = f1_score(y_train, y_pred_train)
    f1score_test = f1_score(y_test, y_pred_test)
    
    accuracytrain.append(accuracy_train)
    accuracytest.append(accuracy_test)
    
    recalltrain.append(recall_train)
    recalltest.append(recall_test)
    
    precisiontrain.append(precision_train)
    precisiontest.append(precision_test)
    
    f1scoretrain.append(f1score_train)
    f1scoretest.append(f1score_test)

And

clf =  sgd_LogisticRegression = SGDClassifier(loss='log', n_jobs = -1, warm_start = True, class_weight = "balanced")

scores = cross_val_score(clf, X, y, cv = KF, scoring = "accuracy")

I am getting vastly different results. The first code was to simply show the cv validation training while I explained in my notebook what I was doing.

But I cannot figure out why to these two snippets of code provide different results. Any ideas?

Thank you!