Closed simonbowly closed 1 year ago
Looks like we are leaning towards:
In 7eb8382 I added optional overloading for bipartite matching, with tests that are skipped if networkx is not installed. faff520 sets up a separate CI step for tests that need required dependencies, so we can be sure not to accidentally require a non-core dependency.
Closing this, we have our final list:
After we have a good collection of models, we should think about which dependencies are the default. Should the user have easy ways (via extras) to install specific dependencies for graphs, machine learning, OR models, without getting everything under the sun with the default install?