This repository contains all the code and document files from the Bayesian Symbolic Regression internship report. It is possible to audit the development of this project through the git history of this repository (e.g. meeting minutes, old analyses, discarded plans).

All of our results (i.e. raw output, intermediary results and final results) are available in the following Open Science Framework repository: https://osf.io/p8bg5/. Our Bayesian symbolic regression program is hosted in the following repository: https://github.com/cobac/BayesianSR.jl

The repository is structured in the following way:

products/

Contains the research proposal and the internship report, as well as feedback for different versions of both documents.

scripts/

Contains the analysis scripts and folder structure for all the analysis of the report. Each sub-folder contains code for different models, and the code of the parent repository integrates all intermediary results to generate the figures that are included in the report.

• All code is written in the Julia programming language. To run a script instantiate the Manifest.toml of the corresponding folder to guarantee that the correct versions of all dependencies are used (do not update the packages!). More info about reproducing Julia projects at https://pkgdocs.julialang.org/v1/environments/#Using-someone-else’s-project .

All of our results can be replicated if each xxx.jl file is run in order for each sub-folder in scripts/, and then the final analysis scripts. Alternatively, it is possible to download from the OSF repository intermediary outputs and only partially audit our results. Finally, it is also possible to download all output.jl2 files from the OSF repository and further analyze our results.