How I Set Up Projects

Notes on keeping everything organised with a clear work flow and file organisation.

Structure

• create project root folder
• open in software

project/

├── R/
├── data/
│   ├── raw/        # source data (never edited)
│   ├── versions/   # timestamped snapshots
│   ├── processed/  # cleaned / merged
│   └── final/      # datasets used for analysis / sharing
├── outputs/        # figures, tables, model objects
├── docs/           # protocols, codebooks, notes
├── renv.lock

notes: - root folder always treated as root - no setwd() - all paths relative - never change raw data

Environment

initialise on first open:

# create renv and lockfile in root folder 
renv::init()

# install packages 
install.packages(c("tidyverse", "here", "renv", "usethis", "devtools"))

# snapshot to save in lockfile for reproducibility and collaboration
renv::snapshot()

notes: - renv.lockfile tracks package versions - collaborators use renv::restore() to get same environment

Paths

• anchor scripts using here package
• avoid using absolute paths

# shows where file is saved relative to root 
here::i_am("R/01_data_processing.R")

# import data using relative paths
readr::read_csv(here::here("data/raw/survey_data_raw.csv"))

Secrets

• use usethis to create .Renviron file in root folder and add secrets in there
• for example: API keys, database credentials etc.
• github will ignore this file so its not shared with collaborators

usethis::edit_r_environ()

• after running, edit .Renviron file
• format specifically:

API_KEY="your_api_key_here"
DB_USERNAME="your_db_username_here"
DB_PASSWORD="your_db_password_here"

after saving the .Renviron file, access the secrets in your R code:

api_key <- Sys.getenv("API_KEY")
db_username <- Sys.getenv("DB_USERNAME")
db_password <- Sys.getenv("DB_PASSWORD")