Replication Code: Connecting Leaves to the Forest

---

Academic project examining the utility of using Random Forest models for multiple imputation with chained equations in political science.

Code

• helper.py: Extra user-defined functions to help with some of the calculations done in the paper.
• helper.R: Extra user-defined functions to help with some of the calculations done in the paper.
• dcr_rf_imputation_draft_spring_2023.qmd: Compile this file to replicate the paper.
  • For full replication of the paper, change the yaml meta data option params.reproduce to "true" instead of false.
    • Warning: This will be really computationally intensive.
• ols.stan: stan model used to fit regressions in paper to measure bias.
• ols.rds: stored stan object.
• extensions/: useful partials for the formatting of the manuscript.

Drafts

• archived/: older drafts
• dcr_rf_imputation_fall_2022.pdf: older draft
• dcr_rf_imputation_draft_spring_2023.qmd: current draft

renv and .venv

• renv/: produced with renv::snapshot(). Documents packages used in project, their dependencies, and the versions for all of that.
• .venv/: produced with poetry. Documents python packages used in the project, their dependencies, and the versions for all of that.
• *.lock: documents environment information for renv and .venv
• .tomal: documents system information for .venv
• .Rprofile: requires a loading of the renv information upon R start up in this working directory

Other stuff

• .gitattributes: a file that helps sort out repository code language composition
• .gitignore: do not submit the following files and directories to github.

Contributors and maintainers

• Damon C. Roberts