US energy consumption

This example is based on the Sankey diagrams of US energy consumption from the Lawrence Livermore National Laboratory (thanks to John Muth for the suggestion and transcribing the data). We jump straight to the final result – for more explanation of the steps and concepts, see the tutorials.

[1]:
from floweaver import *

Load the dataset:

[2]:
dataset = Dataset.from_csv('us-energy-consumption.csv',
                           dim_process_filename='us-energy-consumption-processes.csv')

This defines the order the nodes appear in:

[3]:
sources = ['Solar', 'Nuclear', 'Hydro', 'Wind', 'Geothermal',
           'Natural_Gas', 'Coal', 'Biomass', 'Petroleum']

uses = ['Residential', 'Commercial', 'Industrial', 'Transportation']

Now define the Sankey diagram definition.

[4]:
nodes = {
    'sources': ProcessGroup('type == "source"', Partition.Simple('process', sources), title='Sources'),
    'imports': ProcessGroup(['Net_Electricity_Import'], title='Net electricity imports'),
    'electricity': ProcessGroup(['Electricity_Generation'], title='Electricity Generation'),
    'uses': ProcessGroup('type == "use"', partition=Partition.Simple('process', uses)),

    'energy_services': ProcessGroup(['Energy_Services'], title='Energy services'),
    'rejected': ProcessGroup(['Rejected_Energy'], title='Rejected energy'),

    'direct_use': Waypoint(Partition.Simple('source', [
        # This is a hack to hide the labels of the partition, there should be a better way...
        (' '*i, [k]) for i, k in enumerate(sources)
    ])),
}

ordering = [
    [[], ['sources'], []],
    [['imports'], ['electricity', 'direct_use'], []],
    [[], ['uses'], []],
    [[], ['rejected', 'energy_services'], []]
]

bundles = [
    Bundle('sources', 'electricity'),
    Bundle('sources', 'uses', waypoints=['direct_use']),
    Bundle('electricity', 'uses'),
    Bundle('imports', 'uses'),
    Bundle('uses', 'energy_services'),
    Bundle('uses', 'rejected'),
    Bundle('electricity', 'rejected'),
]

Define the colours to roughly imitate the original Sankey diagram:

[5]:
palette = {
    'Solar': 'gold',
    'Nuclear': 'red',
    'Hydro': 'blue',
    'Wind': 'purple',
    'Geothermal': 'brown',
    'Natural_Gas': 'steelblue',
    'Coal': 'black',
    'Biomass': 'lightgreen',
    'Petroleum': 'green',
    'Electricity': 'orange',
    'Rejected energy': 'lightgrey',
    'Energy services': 'dimgrey',
}

And here’s the result!

[6]:
sdd = SankeyDefinition(nodes, bundles, ordering,
                       flow_partition=dataset.partition('type'))
weave(sdd, dataset, palette=palette) \
    .to_widget(width=700, height=450, margins=dict(left=100, right=120), debugging=True)
[ ]: