network visualizations: splot + spaghetti

[jGaboardi]

This issue documents a conversation starter on gitter.

I have been considering how best to integrate spaghetti visualizations into splot. If you check out the Tutorials, all of them use some visualization functions that could be generalized (I think) without too much difficulty (but maybe I am completely wrong). The Network-constrained spatial autocorrelation tutorial directly utilizes splot.

Sometime soon I will add a list of functions found within the spaghetti tutorials that we can look at more closely to decide (1) which ones can be combined; and (2) which ones may not be suitable for splot.

[slumnitz]

Hi James, thank you for initiating this! There are multiple options. We can pack spaghetti specific visualisations into a splot.spaghetti namespace, similar to esda an libpysal. We can also pack visualisations that are potentially useful for a lot of different applications into splot.mapping and then i.e. create spaghetti specific visualisations with high-level wrappers in splot.spaghetti.

[slumnitz]

This issue links to spaghetti review comments: pysal/spaghetti#581

[slumnitz]

Discussion started in pysal/spaghetti#603 about how to best design API for spaghetti plotting functionality. @jGaboardi thoughts on how to link a .plot() method include:

It would be nice to have it on the Network object, but that raises some questions:

1. The .plot() method would have to use element_as_gdf to accommodate plotting functionality
2. The point observations associated with the Network object are technically their own class (PointPattern), but the exist > within an attribute of Network (e.g. Network.pointpatterns).