R projects and working directories

Nicky Wakim

Why do we need R Projects and the here package?

Answer: Reproducibility!

  • Research data and code can reach the same results regardless of who is running the code

    • This can also refer to future or past you!

  • We want to set up our work so the entire folder can be moved around and work in its new location

What is a working directory?

Working directory

The working directory is the folder R is currently “looking at.” When you want to locate a file, it is the starting point for all file paths in your session.

  • The following will show me my current working directory
getwd()
1
Function to look at the current working directory, which is where my .qmd file lives 
[1] "/Users/wakim/Library/CloudStorage/OneDrive-OregonHealth&ScienceUniversity/Teaching/Classes/PUBH_523_26Su/PUBH_523_26Su_site/lessons/10_Rproj_wds"
  • Any relative path starts from the current working directory
read_rds("../data_lessons/hrs_data.rds")
1
If I want to go up a folder, I can use .. to go up one level, and then I can access the data_lessons folder and the hrs_data.rds file.
  • Note that the absolute file path is /Users/wakim/Library/CloudStorage/OneDrive-OregonHealth&ScienceUniversity/Teaching/Classes/PUBH_523_26Su/PUBH_523_26Su_site/lessons/ data_lessons/hrs_data.rds

The problem with working directories

  • You can set the working directory by hand, but you should NOT!

  • When you set a wd manually, it will break if you move the folder or if you are on a different computer

setwd("/Users/jane/Desktop/my_project")

setwd("C:/Users/jane/Documents/school/BIOL401/my_project")
1
Breaks on any other computer
2
Also breaks if you move the folder
  • When you use qmds, running code in the console and rendering result in two different working directories
    • Let’s walk through an example!

R projects and the here package can help fix both of these issues!

What is an R Project?

R Project

An R Project is a file (.Rproj) that sits in a folder and tells RStudio: “This is where I want my working directory to be set automatically.”

I can add an R Project to my PUBH_523_26Su folder. The R project will show up as PUBH_523_26Su.Rproj.

PUBH_523_26Su/ 
├── PUBH_523_26Su.Rproj
├── admin/
├── data_lessons/
├── data_practice/
├── data_project/
│   ├── data_raw/  
│   └── data_processed/    
├── lessons/
├── practice/
├── project/
└── README.txt

When you open the .Rproj file, RStudio sets the working directory to that folder — every time, on any computer.

Let’s create an R Project!

  • You can create a new project and folder
    • File → New Project → New Directory → New Project
    • Give the project a name, choose where to save it, and click Create Project

 

  • You can also create a project from an existing folder
    • File → New Project → Existing Directory
    • Then you’ll browse for the folder you want to create the project in

 

Let’s see an example!

Opening and using an R Project

  • When working with files in an R Project, you want to open and work within the R project

 

  • You can open an R project two ways:
    • Click on the .Rproj file in your folder
    • Open RStudio and select the project from the top-right corner of the window

 

  • Once a project is open, we can work in our files and do not need to set a working directory manually
    • R Studio will automatically set the working directory to the project folder

Opening a script/qmd directly (by double-clicking it) does not activate the project or set the working directory

We still run into some issues with R Projects

We haven’t fixed this issue yet:

  • When you use qmds, running code in the console and rendering result in two different working directories
    • Let’s walk through an example!

We need the here package to fix this issue!

The here package

  • For scripts in subfolders, use here() to build paths from the project root

  • The here() function makes your code start at the R project folder whether we are working in the console or rendering a qmd document

    • Output for here() returns same thing in console and on render
    pacman::p_load(here)
here()
    1
    Need to load the here package to use the here() function
    2
    The here() function returns the path to the project root folder, regardless of where the script lives 
    [1] "/Users/wakim/Library/CloudStorage/OneDrive-OregonHealth&ScienceUniversity/Teaching/Classes/PUBH_523_26Su/PUBH_523_26Su_site"

Now we can navigate to our files in a robust way

  • We can use here() to navigate to files in our project folder
  • We only need to specify the relative path from the project root folder
  • We can input the relative path two different ways:
    • Using a comma to separate the folder/file names (better practice bc it is platform agnostic)
    • Using a forward (or backward on Windows) slash to separate the folder/file names
here("lessons", "01_Welcome")
[1] "/Users/wakim/Library/CloudStorage/OneDrive-OregonHealth&ScienceUniversity/Teaching/Classes/PUBH_523_26Su/PUBH_523_26Su_site/lessons/01_Welcome"
here("images", "PUBH_R.png")
[1] "/Users/wakim/Library/CloudStorage/OneDrive-OregonHealth&ScienceUniversity/Teaching/Classes/PUBH_523_26Su/PUBH_523_26Su_site/images/PUBH_R.png"
here("images/PUBH_R.png")
[1] "/Users/wakim/Library/CloudStorage/OneDrive-OregonHealth&ScienceUniversity/Teaching/Classes/PUBH_523_26Su/PUBH_523_26Su_site/images/PUBH_R.png"

Wrap-up

  • Create one R Project per analysis

 

  • Open work by clicking the .Rproj file or opening in R Studio

 

  • Use relative paths: no setwd(), ever

 

  • Use here() to make relative file paths

R Projects \(+\) here() \(+\) good folder structure

\[=\]

analysis that runs reliably anywhere, for anyone