I am trying to make a model using xgboost, especially with early stopping in the part of cross validation.
And im trying to compare multiples of models, xgboost and the others.
But facing with an error.
The version of xgboost on my local is 2.1.0.

Here’s the codes.

def cross_validate_score(model, data: pd.DataFrame, cv=None, test_data: pd.DataFrame = None, label: str = 'Response', include_original: bool = True, original_data: pd.DataFrame = None):
    if cv is None:
        cv = StratifiedKFold(n_splits=5, shuffle=True, random_state=42)
    if test_data is None:
        raise ValueError("test_data must be provided")
    
    X = data.copy()
    y = X.pop(label)
    
    val_predictions = np.zeros(len(X))
    test_predictions = np.zeros(len(test_data))
    train_scores, val_scores = [], []
    
    for fold, (train_idx, val_idx) in enumerate(cv.split(X, y)):
        X_train = X.iloc[train_idx].reset_index(drop=True)
        y_train = y.iloc[train_idx].reset_index(drop=True)
        X_val = X.iloc[val_idx].reset_index(drop=True)
        y_val = y.iloc[val_idx].reset_index(drop=True)
        
        if include_original:
            if original_data is None:
                raise ValueError("original_data must be provided when include_original is True")
            X_train = pd.concat([original_data.drop(label, axis=1), X_train]).reset_index(drop=True)
            y_train = pd.concat([original_data[label], y_train]).reset_index(drop=True)
        
        model_cloned = clone(model)  
        
        if isinstance(model_cloned, XGBClassifier): 
            eval_set = [(X_val, y_val)]
            model_cloned.fit(X_train, y_train, eval_set=eval_set, early_stopping_rounds=50, verbose=False) # I think this is the main reason of the error. 
            best_iteration = model_cloned.best_iteration
            train_preds_proba = model_cloned.predict_proba(X_train, iteration_range=(0, best_iteration))[:, 1]
            val_preds_proba = model_cloned.predict_proba(X_val, iteration_range=(0, best_iteration))[:, 1]
            test_preds_proba = model_cloned.predict_proba(test_data, iteration_range=(0, best_iteration))[:, 1]

        elif isinstance(model_cloned, LGBMClassifier):
            eval_set = [(X_val, y_val)]
            model_cloned.fit(X_train, y_train, eval_set=eval_set, eval_metric='auc', callbacks=[early_stopping(50)])
            best_iteration = model_cloned.best_iteration_
            train_preds_proba = model_cloned.predict_proba(X_train, num_iteration=best_iteration)[:, 1]
            val_preds_proba = model_cloned.predict_proba(X_val, num_iteration=best_iteration)[:, 1]
            test_preds_proba = model_cloned.predict_proba(test_data, num_iteration=best_iteration)[:, 1]
            
        else:
            raise ValueError("Model type not supported for early stopping.")
        
        val_predictions[val_idx] = val_preds_proba 
        train_scores.append(roc_auc_score(y_train, train_preds_proba))
        val_scores.append(roc_auc_score(y_val, val_preds_proba)) 
        
        print(f'Fold {fold}: {val_scores[-1]:.5f}')
        
        test_predictions += test_preds_proba / cv.get_n_splits() 
    
    print(f'Val Score: {np.mean(val_scores):.7f} ± {np.std(val_scores):.7f} | Train Score: {np.mean(train_scores):.7f} ± {np.std(train_scores):.7f} | {label}')
    
    return val_scores, val_predictions, test_predictions



cv_summary, oof_predictions, test_predictions = pd.DataFrame(), pd.DataFrame(), pd.DataFrame()


xgb_params = {
    'n_estimators': 10000, 
    'eta': 0.05,
    'alpha':  0.2545607592482198,
    'subsample': 0.8388163485383147, 
    'colsample_bytree': 0.2732499701466825, 
    'max_depth': 16,
    'min_child_weight': 5,
    'gamma': 0.0017688666476104672,
    'eval_metric': 'auc',
    'max_bin': 262143, #a weird max_bin, for reference: https://www.kaggle.com/competitions/playground-series-s4e7/discussion/516265   
    'tree_method': 'gpu_hist',
}


xgb_tuned = XGBClassifier(**xgb_params, random_state=42)


%%time 
cv_summary['xgb'], oof_predictions['xgb'], test_predictions['xgb'] = cross_validate_score(xgb_tuned, data=train_df, test_data=test_df, include_original=False)

And the error is like this.

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
File <timed exec>:1

Cell In[58], line 35
     33 if isinstance(model_cloned, XGBClassifier): 
     34     eval_set = [(X_val, y_val)]
---> 35     model_cloned.fit(X_train, y_train, eval_set=eval_set, early_stopping_rounds=50, verbose=False) # 조기 종료 설정
     36     best_iteration = model_cloned.best_iteration
     37     train_preds_proba = model_cloned.predict_proba(X_train, iteration_range=(0, best_iteration))[:, 1]

File c:Usersdxschoolanaconda3Libsite-packagesxgboostcore.py:726, in require_keyword_args.<locals>.throw_if.<locals>.inner_f(*args, **kwargs)
    724 for k, arg in zip(sig.parameters, args):
    725     kwargs[k] = arg
--> 726 return func(**kwargs)

TypeError: XGBClassifier.fit() got an unexpected keyword argument 'early_stopping_rounds'

Is there anybody who had faced with this problem, exactly same to mine?
I really need you guy’s help!!!