My objective here is to reproduce the results of Diamant et al. (2024), a study on the observed changes in dolphin vocalizations in the presence of underwater radiated noise from ships.
Diamant, R., Testolin, A., Shachar, I. et al. Observational study on the non-linear response of dolphins to the presence of vessels. Sci Rep 14, 6062 (2024). https://doi.org/10.1038/s41598-024-56654-6
The study found that a machine learning-based classifier could be trained on whistle features to discriminate between whistles emitted in the presence of URN versus those that were not.
Using the feature file from the original study, I will attempt to replicate the logic and methodology of their model fitting, validating, and testing.
I am implementing this using R and the tools I am most familiar with. This is in contrast to the tools used by the original authors, which were all MATLAB-based.
Along the way, I am striving to make my project as reproducible as possible, using purpose-built libraries, and GitHub flow to enhance reproducibility at every step of the way:
renv: Making my environment reproducible, generatesrenv/andrenv.lockfiletargets: Used to write the script_targets.Rthat sets up the pipeline that I use to process data, fit the models, and generate results.- Also I am trying to capture as much analysis as possible in a Quarto notebook
notebook.qmdthat I am publishing online
Notebook with the reproduced figures is accessible here
Looking for feedback from knowledgeable individuals to improve my learning from this project. Thanks.
Funded through the Association for the Sciences of Limnology and Oceanography (ASLO) / Limnology and Oceanography Research Exchange (LOREX). More information about the program here.
Thanks also to Roee and everyone at the Acoustic Navigation Laboratory in the Hatter Department of Marine Technologies, University of Haifa. Lab website