Explore interoperability opportunities

[maxhully]

In a perfect world, we could write our metrics so that they would work with scalar values, vectors (like numpy arrays), pandas Seriess, and our Partitions. What kind of API changes would be required for that to be possible?

[maxhully]

I did some work in this direction, writing a small graph/graph partition library that uses SciPy sparse matrices for holding adjacency information and pandas DataFrames for node and edge data: https://github.com/maxhully/partitions

[pjrule]

This direction seems to have been abandoned. Some of the functions in GerryTools provide metrics that are not strongly coupled to the Partitionobject and can read DataFrames instead.