Research Interactive Visualisation Techniques

[CharlotFranssen]

• Search possible libraries -> pyvis (as the other group used in their supervisory meeting)
• Create summary on findings
• Research Ora

[imbanunes]

Packages/libraries options:

1. pyvis (best for directed graphs in general, apart from acyclic) https://pyvis.readthedocs.io/en/latest/index.html

   • cons: no information sidebar, scales worse, more complex coding
   • pros: pop-up info on node/edge, better filtering functionality (menu available), supports dragging and dropping nodes, allows searching for nodes or edges based on their attribute values, better customisation experience (interactive dashboard)

2. ipysigma (best for undirected graphs) https://github.com/medialab/ipysigma (jupyter widget)

   • cons: some interactiveness is lost, allows selection but not hiding nodes, only allows for searching on the node identifier value, unusable for heterogenous directed graphs (edges overlap)
   • pros: convenient interactive information sidebar, scales better, less cumbersome code, better for undirected homogeneous graphs

3. others python options: networkx, dash, plotly

4. R options (not nearly as sophisticated): the favourite seems to be visNetwork (which has an interface with igraph)

ORA (Organizational Risk Analyzer):

• organized by the questions the user asks, not the metrics. As such, common workflows are automated and all metrics relevant to a particular question are automatically included in a report providing the network answer to the question of interest.
• size limits – over 1 million nodes in batch (free lite version is however limited)
• programming language – Java GUI and C++ backend
• supports high dimensional network data, analysis on very large network data sets; is interoperable with a large number of other tools; has been designed to reduce training time and effort on the part of the user; provides quite some guidance tools;

[WouterMulder9]

Looks good!