Work Environment
Before you start: running R commands.
Setup: Open your RStudio project; this unit supplies a practice file for Data/ and explains relative paths.
Your check: Reopen a project and read the supplied Data/example.csv using a relative path; verify three rows and two columns. Practise in the unit summary.
“Data is the new oil.” — Clive Humby
What does workspace mean?
In R, the workspace contains the objects you have created in the global environment during your session, such as vectors, data frames, and functions. RStudio shows these objects in the Environment pane.
The working directory is a folder on your computer. R uses it as the starting point for relative file paths. The files in a folder and the R objects in your workspace are different things: reading a CSV file creates an object in R, while the CSV file remains on disk.
An RStudio project organizes your work in a project folder and stores project settings in an .Rproj file. Opening the project sets the working directory to that folder. Use this as your starting point for the course’s file examples.
For a reminder of the panes, revisit the Unit 01 interface guide. Here the new distinction is between objects in memory and files on disk: identify one of each in your own project before continuing.
Start with an RStudio project
If you already followed First Steps, open that project’s .Rproj file. Otherwise, choose File > New Project > New Directory > New Project, name the new directory base-r-practice, choose a location you can find again and click Create Project. If your course already provides a project, open that one instead.
Create a Data/ folder inside the project folder. Download example.csv and save it there, keeping its name. This is the same small practice file used in R Syntax, with comma-separated columns and decimal points. Save your .R script beside the .Rproj file.
base-r-practice/
├── base-r-practice.Rproj
├── first-import.R
└── Data/
└── example.csv
Run these commands from the project folder:
getwd()
file.exists("Data/example.csv")
trees <- read.table(file = "Data/example.csv", header = TRUE, sep = ",")
dim(trees)
getwd() should show your own project folder. file.exists() should return TRUE, and dim(trees) should return 3 2: three rows and two columns. If the file check is FALSE, reopen the correct .Rproj and check the filename and folder, including capitalization and an accidental extra extension. Keep the working directory at the project folder; Data/ in the path already selects the data folder.
Check portability: save your script, close and reopen the project via its .Rproj, then run the script from the top. The same relative path should work. Moving or sharing the whole project folder keeps that path usable as long as the internal arrangement stays the same. Create Output/ inside the project folder when a later lesson asks you to save results.
The remaining sections explain larger folder arrangements and alternative path choices. They use illustrative paths such as C:/geomoer/Project1/; adapt them to your computer if you try them. The supplied-file workflow above is the recommended starting route.
Project directory structure
If you have a project with many data files, analysis scripts or other it is highly recommended that you create a directory structure under your main project folder to store and organize all these files. For example, you might have a project folder under C:/geomoer/Project1, whose directory structure might look like this
The key elements of a project folder include:
- a
Data/folder - a
Process/folder and - an
Output/folder.
Keep unchanged downloaded inputs in Data/. Save prepared tables and other generated results under different filenames in Output/, as in Unit 06, so that running a script does not overwrite its input.
Process/ is an optional arrangement for larger projects. It can contain scripts, intermediate working files and subfolders for figures or maps while you develop an analysis. In a small practice project, keeping the .R script beside the .Rproj file, as above, is sufficient.
Use Output/ for generated exports, plots, reports and presentations, for example CSV, RDS, PNG, PDF or HTML files. An .Rmd is an editable source document; keep it with your scripts and place its rendered report in the chosen output location. You do not need all these formats or subfolders for every task. Keep the working directory at the project root and use the same folder capitalization in your paths.
Understanding the directory structures
The following reference example uses an illustrative file named data.csv, rather than the supplied example.csv. Its code and computer-specific paths are templates; the deliberately failing import explains why the starting folder matters.
R resolves a relative file path from the current working directory. Saving or opening an R script in another folder does not by itself change the working directory.
In this example, user Geomoer has a project folder called Project1 containing a Data/ folder and a Process/ folder. The file data.csv is in Data/, and the R session’s working directory is C:/geomoer/Project1/Process. Running the following line produces an error:
data <- read.csv("data.csv")
Error in file(file, "rt"): cannot open the connection
R looks for data.csv in Process/, where it does not exist.
An R session’s working directory can be verified by typing the following command:
getwd()
[1] "C:/geomoer/Project1/Process"
The working directory is also the starting point for relative paths when writing files. A path such as Data/data.csv is still relative; including a folder name does not make it an absolute path.
Besides opening a project and using a path relative to its root as above, there are two alternatives to resolve this illustrative problem.
The first is to set the working directory to the folder that contains data.csv using the setwd() function, then read the file:
setwd("C:/geomoer/Project1/Data")
dat <- read.csv("data.csv")
The second is to specify an absolute path in the read.csv() call. This works regardless of the current working directory:
dat <- read.csv("C:/geomoer/Project1/Data/data.csv")
Hint: Pay attention to the quotation marks.
Relative and absolute paths
Hard-coded absolute paths make it difficult to share the project folder with someone else who may place it under a different folder such as D:/Uni/R/Test/Project1/.
The user would need to modify paths that refer to C:/geomoer/Project1/. A relative path can avoid this. For the following example, first reset the working directory to Process/ if you changed it above:
setwd("C:/geomoer/Project1/Process")
dat <- read.csv("../Data/data.csv")
The two dots .. tell R to move up the directory hierarchy relative to the current working directory. So in our working example, ../ tells R to move out of the Process/ folder and up into the Project1/ folder. The relative path ../Data/data.csv tells R to move out of the Process/ directory and over into the Data/ directory before attempting to read the contents of the data.csv data file.
After moving the project, the same relative path works as long as Data/ and Process/ keep the same arrangement and the working directory is again Process/. You still need to select that working directory at its new location. An RStudio project can help establish a consistent starting directory, as shown below.
Absolute and relative paths
An absolute (or full) path points to the same location in a file system, regardless of the current working directory. To do that, it must include the root directory. For example: C:/geomoer/Project1/Data/data.csv
By contrast, a relative path starts from the current working directory. From C:/geomoer/Project1/Process, use ../Data/data.csv. From C:/geomoer/Project1, use Data/data.csv.
Important Note to Windows Users:
R gets confused if you use a path in your code like:
c:\mydocuments\myfile.txt
This is because R sees "\" as an escape character. Instead, use:
c:\\my documents\\myfile.txt
or
c:/mydocuments/myfile.txt
Either will work. We use the second convention throughout this platform.
Further project details
When you open an .Rproj file in RStudio, the working directory is set to the folder containing that file. For example, if Project1.Rproj is in C:/geomoer/Project1/, use a path relative to this project folder:
dat <- read.csv("Data/data.csv")
This also works after moving the whole project folder and opening the .Rproj file at its new location. Keep the working directory at the project folder while running these examples. If you change it with setwd(), relative paths are resolved from the new working directory. Use getwd() to check your current location.
The .Rproj file stores project settings; your scripts and data remain separate files in the project folder. See the RStudio Projects documentation for more information.
Now move to the packages…