Fig. 3.

Performance comparison of models. a) Performance comparison of models trained on seven types of phylogenetic trees. Each bar depicts the relative error observed when testing trees of the same type as those used in training. The horizontal lines above the boxes denote the median relative error when testing the Baseline-Model and Dated Baseline-Model with Dated Resolved Trees, as well as the Genetic Baseline-Model with Genetic Resolved Trees. Models trained using poorly resolved phylogenetic trees (i.e. Dated Resolved, Dated Polytomous, Genetic Resolved, and Genetic Polytomous) are highlighted in bold. b) Performance comparison of models tested using poorly resolved phylogenetic trees. “Baseline-Poor” represents the evaluation of the Baseline-Model tested using Dated Resolved Trees. “Dated Baseline-Poor” indicates the assessment of the Dated Baseline-Model with Dated Resolved Trees, while “Genetic Baseline-Poor” reflects the performance of the Genetic Baseline-Model when testing with Genetic Resolved Trees.