import pandas as pd
import numpy as np
pd.options.display.notebook_repr_html = False

# hierarchical indices and columns
index = pd.MultiIndex.from_product([[2014,2016,2018], [1, 2, 3]],
                                   names=['year', 'sample_num'])
columns = pd.MultiIndex.from_product([['Kinshati', 'Ryujinboku', 'Ranpougyoku'], ['width', 'Height']],
                                     names=['Cactus', 'length'])
# mock some data
data = np.round(np.random.randn(9, 6), 1)
data[:, ::2] *= 5
data += 10

# create the DataFrame
cactus_data = pd.DataFrame(data, index=index, columns=columns)
cactus_data
"""
Cactus          Kinshati        Ryujinboku        Ranpougyoku
length             width Height      width Height       width Height
year sample_num
2014 1              11.5   10.4       15.5    9.3        18.5    8.7
     2              20.0    9.8       10.5   10.2        20.0   10.5
     3              11.5   10.4       18.5   11.1         6.0    9.7
2016 1               5.5    9.3       13.0    9.1         6.5    8.4
     2               5.5    9.1        6.0    9.5        10.5   10.1
     3              11.5   10.9        6.0    9.0        10.0    9.0
2018 1              17.5    9.1       14.5   10.7        13.0   10.6
     2               9.5   10.4        1.5   10.6        13.5   10.6
     3               4.5    9.7       12.5    8.9         9.5    9.8
"""_

data_heikin = cactus_data.mean(level='year')
data_heikin
"""
Cactus   Kinshati            Ryujinboku            Ranpougyoku
length      width     Height      width     Height       width     Height
year
2014    14.333333  10.200000  14.833333  10.200000   14.833333   9.633333
2016     7.500000   9.766667   8.333333   9.200000    9.000000   9.166667
2018    10.500000   9.733333   9.500000  10.066667   12.000000  10.333333
"""

data_heikin2 = cactus_data.mean(axis=1,level='length')
data_heikin2
"""
length               width     Height
year sample_num
2014 1           15.166667   9.466667
     2           16.833333  10.166667
     3           12.000000  10.400000
2016 1            8.333333   8.933333
     2            7.333333   9.566667
     3            9.166667   9.633333
2018 1           15.000000  10.133333
     2            8.166667  10.533333
     3            8.833333   9.466667
"""