aumag/ausankey

create sankey diagrams with matplotlib

ausankey

Uses matplotlib to create simple Sankey diagrams flowing only from left to right.

This package is available on PyPi and can be installed via:

pip install ausankey

User documentation for the repository is published via GitHub Pages: https://aumag.github.io/ausankey/

Code documentation by Mkdocs is available here: https://aumag.github.io/ausankey/reference/

Minimal example

import ausankey as sky
import matplotlib.pyplot as plt
import pandas as pd

data = pd.DataFrame([
  ("a",1.0,"ab",2.0,"a",1.0),
  ("a",1.0,"ba",0.8,"ba",0.4),
  ("c",1.5,"cd",0.5,"d",2.0),
  ("b",0.5,"ba",0.8,"ba",0.4),
  ("b",0.5,"ab",0.8,"a",1.0),
  ("d",2.0,"cd",0.4,"d",1.0),
  ("e",1.0,"e",1.0,"e",3.0),
])

plt.figure()
sky.sankey(
  data,
  sort      = “top”,
  titles    = ["Stage 1","Stage 2","Stage 3"],
  valign    = "center",
)
plt.show()

Requirements

• Python 3.x
• matplotlib
• numpy
• pandas

ausankey's Issues

Groups of labels

It might be desirable to sort and colour labels in groups rather than treating them all as independent. Not sure if data frames might allow this in a sensible way for data entry

Add pypi packager

Last step to setting up the repo properly (at least for this round of updates) is to get listed in pypi

not sure how to do this yet! Will need to look around

Boundaries around flows

Switching to the gradient model has meant that we lost an outline around each flow. It would be good to (optionally) bring this back by plotting another line over the boundary

Rows need to be able to start/stop

I.E., a label should be able to split into two.
This would need syntax like

(“A”,1,”B”,1,None,None),
(None,None,”B”,0.5,”C”,0.5),
(None,None,”B”,0.5,”D”,0.5),

Or similar…

Top to bottom rather than left to right?

It would be relatively sensible to allow a vertical design instead of a horizontal one. Might not even need that much extra branching code.

Options to customise text

Need a way to pass through text formatting options for titles and labels (separately).