NetIn - 2.0.0a2

What's Changed

• Fix to link node IDs wenn exporting graph to networkx by @mannbach in #30
• Rename variables to align with netin-1.0 by @mannbach in #28 and #27
• Fixed bugs and updated variable names of PATCHModel

New Contributors

• @samuel-mg made their first contribution in #24

Full Changelog: netin-1.0.7...netin-2.0.0a2

netin-2.0.0a1

What's Changed

This update presents a major restructure of the code base to improve simulation performance and extendability.

• Components are now split into classes. For instance, netin/graphs/graph.py defines a Graph class that represents networks created by Models which are defined in netin/models/model.py. Realizations of models, such as the PAHModel in netin/models/pah_model inherit from this abstract Model-class and implement the actual simulation logic by combining LinkFormationMechanisms, as implemented in netin/link_formation_mechanisms. Each of these, and other, components can be extended by the user to create their own model variants.
• LinkFormationMechanisms utilize vectorization when computing the probabilities to connect to a node. This is achieved by introducing the NodeVector-class (netin/graphs/node_vector.py) which wraps numpy.ndarrays to keep track of node-based attributes like connection probabilities, degrees or minority status. This vectorization achieves a significant speedup compared to the older implementation.

An extensive patchlog will follow for the production release.

New Contributors

• @samuel-mg made their first contribution in #24

Full Changelog: netin-1.0.7...netin-2.0.0a1

NetIn 1.0.7

Includes

1. Sampling
   1. Random Nodes
   2. Random Edges
   3. Random Neighbor
   4. DegreeRank (desc and asc)
   5. DegreeGroupRank (desc and asc)
   6. PartialCrawls
2. Ranking inequalities (stats in netin.stats.ranking, and plots in netin.viz.handlers
   1. Inequality (Gini in top-k)
   2. Inequity (Mean error to parity of fraction of minorities in top-k)
   3. Disparity (inequality vs inequity)
3. Examples in examples/notebooks
   1. sampling
   2. ranking
4. Documentation using sphinx

What's Changed

• Add auto-generated documentation to the project by @mannbach in #2
• Add logo to documentation and README by @mannbach in #5

New Contributors

• @mannbach made their first contribution in #2

Full Changelog: netin-1.0.5...netin-1.0.7

NetIn 1.0.5

Alpha version, includes:

[MODELS]

1. Undirected
   1.1. PA (Barabasi)
   1.2. PAH (Barabasi with Homophily)
   1.3. PATC (Barabasi with Triadic Closure)
   1.4. PATCH (Barabasi with Triadic Closure and Homophily)
2. Directed
   2.1. DH (Directed with Homophily)
   2.2. DPA (Directed with Preferential Attachment)
   2.3. DPAH (Directed with Preferential Attachment and Homophily)

[EXAMPLES]
In the folder examples/ you can find python scripts (.py) and notebooks (.ipynb) with examples on how to create graphs, plots, and ranking analysis.

Full Changelog: https://github.com/CSHVienna/NetworkInequalities/commits/netin-1.0.5