xgb_baseline.py 文件源码

def load_data():

    train_data = pd.read_csv(os.path.join(data_folder, 'train.csv'), delimiter=';', skip_blank_lines=True)
    test_data = pd.read_csv(os.path.join(data_folder, 'test.csv'), delimiter=';', skip_blank_lines=True,
                            na_values='None')

    ntrain = train_data.shape[0]
    ntest = test_data.shape[0]

    print('ntrain={}'.format(ntrain))
    print('ntest={}'.format(ntest))

    y_train = train_data['cardio'].values

    # --------------------------------------------------------------

    x_train = train_data.drop(["id", "cardio"], axis=1)
    x_test = test_data.drop(["id"], axis=1)

    x_test.replace('None', np.nan)

    return (x_train,y_train,x_test)

# ---------------------------------------------------------------------