new_vote.py

def vote(estimators, threshold, X_train, y_train, X_val, y_val, X_test, score_name):
    
    predict_list = []
    estimator_list = []
    
    for estimator_name in estimators:  
        try:
            params, val_score, timestamp = get_params(estimator_name)
            if val_score > threshold:
                print('Ensemble add %s' % estimator_name)
                estimator_list.append(estimator_name)
                if estimator_name == 'keras':
                    clf = keras_model(**params)
                    model_path = os.path.join('saves/', timestamp + '.json')
                    weights_path = os.path.join('saves/', timestamp + '.h5')
                    clf_pred = clf.predict_proba(X_val, model_path, weights_path)
                    predict_list.append(clf_pred)                  
                else:
                    if estimator_name == 'xgb':
                        clf = XGBClassifier(nthreads=-1, **params)                        
                    elif estimator_name == 'log':
                        clf = LogisticRegression(**params)
                    elif estimator_name == 'knn':
                        clf = KNeighborsClassifier(**params)
                    elif estimator_name == 'rfo':
                        clf = RandomForestClassifier(**params)
                    elif estimator_name == 'ada':
                        clf = AdaBoostClassifier(**params)
                    elif estimator_name == 'ext':
                        clf = ExtraTreesClassifier(**params)
                    elif estimator_name == 'svc':
                        clf = SVC(**params)
                        
                    clf.fit(X_train, y_train)
                    clf_pred = clf.predict_proba(X_val)
                    predict_list.append(clf_pred)
            
        except: FileNotFoundError
 
    y_pred = np.mean(predict_list, axis=0)
    y_pred = np.argmax(y_pred, axis=1)
    return y_pred