Setting Up an R Project

Organize your files so your analysis is reproducible and easy to share

Use RStudio Projects (.Rproj)

An RStudio Project keeps everything for one analysis — scripts, data, and results — together in a single folder. Instead of scattering files across your desktop and downloads, you work inside one self-contained folder that you can copy, back up, or hand to a colleague as a unit.

To create one in RStudio, go to File → New Project → New Directory → New Project, choose a name and a location, and click Create Project. RStudio makes the folder and puts a small .Rproj file inside it (for example my_project.Rproj).

That .Rproj file is what makes projects worthwhile:

Open the project by double-clicking the .Rproj file. RStudio opens with that folder as the working directory — the place R looks for your files.
Because the working directory is set automatically, you never need setwd() with a hard-coded path. This is the single biggest step toward reproducible code.
Everything stays relative to the project folder, so the project works the same on any computer.

Habit to build

Start every new practical by creating a project first, then writing code. A few seconds of setup saves a lot of “R can’t find my file” confusion later.

A recommended folder structure

Inside your project folder, group files by what they are. A simple, reliable layout looks like this:

📁 my_project/
├── 📄 my_project.Rproj    # the project file — open this to start
├── 📁 data/               # raw data files (read-only)
├── 📁 R/                  # your .R analysis scripts (some prefer scripts/)
├── 📁 output/             # generated results, cleaned data
└── 📁 figures/            # saved plots

📁 = folder · 📄 = file

Folder / file	What goes here
`data/`	Original data exactly as you received it. Treat it as read-only — never overwrite the raw files.
`R/` (or `scripts/`)	Your `.R` scripts that import, clean, analyse, and plot.
`output/`	Results your code produces: cleaned datasets, tables, summaries.
`figures/`	Plots you save with `ggsave()` or similar.

Keeping raw data separate and untouched means you can always re-run your analysis from scratch and get the same result. If something goes wrong, your original data is still safe.

This is the layout used on this site

Every dataset on this site lives in a data/ folder and is read with a project-relative path such as read_csv("data/tablet_weights.csv"). When you set up your own project the same way, the examples here work unchanged.

Download a ready-to-run example

Grab this exact layout as a working project: Download the example project (ZIP).

Unzip it, open my_project.Rproj in RStudio, then open R/analysis.R and click Source to run it. Using the tablet_weights.csv dataset, it prints a per-batch summary, writes output/batch_summary.csv, and saves figures/weight_by_batch.png — no setup needed beyond having the tidyverse installed.

Paths and the working directory

Once you work inside a project, always refer to files with project-relative paths — paths that start from the project folder, not from the top of your hard drive.

# Do this — relative to the project folder
tablet_weights <- read_csv("data/tablet_weights.csv")

# Avoid this — an absolute path tied to one computer
tablet_weights <- read_csv("C:/Users/Morten/Documents/practical3/data/tablet_weights.csv")

# Avoid this — changing the working directory by hand
setwd("C:/Users/Morten/Documents/practical3")

Why avoid absolute paths and setwd()

A path like C:/Users/Morten/... only exists on your machine. The moment a classmate, a teacher, or a lab computer opens your script, it breaks. Project-relative paths work everywhere, so your analysis stays reproducible.

File naming conventions

Consistent file names make a project easy to scan and keep scripts in a sensible order.

Use snake_case: lowercase words joined by underscores.
No spaces and no special characters — they cause problems on different systems.
Number scripts when they run in sequence, so they sort in run order:

R/
├── 01_import.R      # read the raw data
├── 02_clean.R       # tidy and compute new columns
├── 03_analysis.R    # statistics and models
└── 04_figures.R     # produce plots

Naming inside your code

These rules are for file and folder names. For naming variables, functions, and columns inside your R code, follow the R Coding Conventions — the same snake_case idea, with units in the name (fill_mass_mg, pct_label).

Reproducibility habits

A few small habits make your work reliable enough that anyone — including you, months later — can reproduce it:

Load packages at the top of every script with library(), so it is clear what the script needs.
Restart R with a clean environment and re-run the whole script from the top before you trust a result. This proves the script works on its own, not because of leftover objects in memory. (In RStudio: Session → Restart R.)
Keep raw data read-only — save cleaned data to output/ instead of overwriting files in data/.