Open whelena opened 1 year ago
Graphs like this would be useful! As it stands, the algorithm relies on the constraints set by tree structures (single-parent per child) in order to arrange the nodes properly. This wouldn't be too disruptive to the existing algorithm if we could consider a single parent the "true" primary parent for the purposes of positioning nodes. Then, we would just draw other secondary arrows (which could be styled using color, linestyle, opacity, etc.) without affecting node positions. This would surely result in some ugly trees, but users would have many options for resolving those issues depending on their particular tree (angle overrides, lengths, opacity, etc.).
Add probability column to input dataframe
Probability column can also be a p-value. Probabilistic edge (connected to a node with multiple parents) should be different in colour or line type compared to 'exact' edges (connecting node with only one parent).
Example:
** labels (0.6, 0.4) do not have to be included