Closed voetberg closed 8 months ago
Should start with the graphs that fail miserably (in the test set) - The more physically interesting sets can be picked later. Produce visuals of the bad classifications with the graph edges (weighted)
Picked out a test set. It's a little more than two.
Selected based picking the highest and lowest recall/precision for each class (excluding showers)
(Only one or two graphs?)