Login / Signup

Complexity is complicated and so too is comparing complexity metrics-A response to Mikula et al. (2018).

William D PearseIgnacio Morales-CastillaLogan S JamesMaxwell FarrellFrédéric BoivinT Jonathan Davies
Published in: Evolution; international journal of organic evolution (2018)
In a recent publication (Pearse et al. 2018b), we explored the macroevolution and macroecology of passerine song using a large citizen science database of bird songs and powerful machine learning tools. Mikula et al. (2018) examine a small subset (<8%) of the data we used, and suggest that our metric of song complexity, the SD of frequency (SDF), does not correlate to other metrics of birdsong complexity, specifically syllable repertoire size and syllable diversity. We comment on the diversity of complexity metrics that exist in the field at present, and, while acknowledging that metrics may differ, outline how this variety allows us to ask more biologically nuanced questions. We see no reason or need for all complexity metrics to be correlated. Since different complexity metrics have been, and will continue to be, used, we outline how metrics could be fairly compared in the future.
Keyphrases
  • machine learning
  • public health
  • big data
  • emergency department
  • electronic health record
  • deep learning