Fig. 9
From: Constructing phylogenetic networks via cherry picking and machine learning
Experimental results for ZODS data. Each point on the horizontal axis corresponds to one instance group. For the graphs on the left, there is one group for each fixed pair \((L,|\mathcal {T}|)\) consisting of 16 instances coming from ZODS networks for each value of \(R\in \{10,20,30\}\). For the graphs on the right, there is one group for each fixed pair (L, R) consisting of 16 instances coming from ZODS networks for each value of \(|\mathcal {T}|\in \{20,50,100\}\). In the top graphs, the height of each bar gives the average of the results it represents over all instances of the group, each scaled by the number of reticulations of the network the instance originated from. The bottom graphs compare the average output of ML within each instance group and the average of the best output given by TrivialRand for each group instance. The shaded areas represent 95% confidence intervals