You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
We need more ways to compare a discovered graph to a ground truth graph.
Describe the solution you'd like
Overlay two graphs using graph viz, and use different edge colors to indicate:
Successes: Edges that are the same in both graphs
Success but wrong direction: Edges that are aligned one way in one graph and aligned the other way in the other graph. Alternatively, edges that are undirected in one graph but not undirected in the other
False positives: Edges that are present in the discovered graph but missing in the ground truth graph
False negatives: Edges that are missing in the discovered graph but present in the ground truth graph
Describe alternatives you've considered
We've proposed structural Hamming distance and confusion matrices. However, graph visualization is more intuitive and it shows you where exactly in the graph you made your errors, which can be much more informative than a statistic.
Additional context
This exists in bnlearn though it is not clear to the user which graph is the graph being evaluated (the discovered graph) and which graph is the reference graph (the ground truth).
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
We need more ways to compare a discovered graph to a ground truth graph.
Describe the solution you'd like
Overlay two graphs using graph viz, and use different edge colors to indicate:
Describe alternatives you've considered
We've proposed structural Hamming distance and confusion matrices. However, graph visualization is more intuitive and it shows you where exactly in the graph you made your errors, which can be much more informative than a statistic.
Additional context
This exists in bnlearn though it is not clear to the user which graph is the graph being evaluated (the discovered graph) and which graph is the reference graph (the ground truth).
The text was updated successfully, but these errors were encountered: