Skip to content

Latest commit

 

History

History
44 lines (31 loc) · 1.74 KB

README.md

File metadata and controls

44 lines (31 loc) · 1.74 KB

Frequency-dependent selection in timeseries data, analysis

This repository contains code and data to reproduce the results and figures of Newberry and Plotkin (2022) Measuring frequency-dependent selection in culture.

The companion source code repository is fdsel.

Dependencies

The analysis relies on the software fdsel, available in a separate repository and R (3.6.1) with packages optparse and ggplot2, as well as a typical unix environment.

The binary fdsel is required to be in the path. Some simulations require OCaml components that must be compiled in the fdsel source repository and the binaries copied here. These are described in comments in the respective run.sh files.

Name analysis

The name analysis is in the names subdirectory. Executing ./run.sh and Rscript plot.R reproduces the analysis. Please note that bootstrap confidence intervals required to reproduce main text figures take several CPU-days to execute on modern CPUs.

Dog popularity analysis

The dog popularity analysis is in the akc subdirectory. Executing ./run.sh and Rscript plot.R reproduces most of the analysis.

The novelty bias grid search (SI Section 5) is extremely computationally intensive and takes dozens of CPU-days to execute. To run this analysis, uncomment the relevant lines from run.sh.

Author

All software was written by Mitchell Newberry mitchell@localpost.io and is (c) Mitchell Newberry 2020-2022. Bug reports and comments are welcome.

Reproductions of public data included in this repository for archival purposes and convenience. Their copyright and licensing terms lie with thier respective authors as described in LICENSE.md, names/inp/PROVENANCE.md and akc/inp/PROVENANCE.md.