feature_table_builder.py 文件源码

def label_feature_producer(start_date, end_date, features, labels):
    labeller = Labeller(start_date, end_date, labels)
    dataset = labeller.get_labels()
    dataset['training_end_date'] = start_date
    dataset['labeling_end_date'] = end_date
    # dataset
    feature_grabber = feature_processor.FeatureGrabber(start_date, engine,config_db, con)
    feature_name_dictionary = {}

    #print dataset
    for feature in features:
            #print feature
        res_training, feature_names_training = feature_grabber.getFeature(feature)
        feature_name_dictionary[feature] = feature_names_training
        res_training.drop_duplicates(inplace = True)
        dataset = pd.merge(dataset, res_training, on = config_db['id_column'],
                           how = 'left')

    return dataset, feature_name_dictionary

    #return None, None