InfoGuides: Software: Learn R: Data Management

Data Management in R - The 3 Options

Although it is not required to use just one of these, it is best to choose one and use it consistently. However, it is useful to be able to read Base R code, as it may be used in tutorials, is necessary for certain packages, and can simplify code in some circumstances.

If you really can't decide, check out this comparison for examples of each.

Scheme	Data Class	Template	Advantages
Base R	data.frame	mydata[rows,columns]	universal, historical
Data Table	data.table	mydata[rows,columns,by]	fastest for Big Data
Tidyverse	tibble	mydata \|> filter(rows) \|> select(columns)	easiest to use

We recommend the Tidyverse for most people who are doing data analysis from the social and health sciences. Data Table can be better for Data Scientists and programmers or those who work with very large data files. See Learn to work with large dataset in R by Analytics Vidhya

Few people use Base R exclusively, but it can be useful to know some notation such as $ and [] referring to object parts (see foundations), especially for unique packages that do not support tidyverse notation.

What is tidy data? See the classic article: Tidy Data (pdf) by Hadley Wickham, uses an older version of tidyr

Tidyverse Tutorials & Documentation

Tidyverse Documentation - Click on individual packages for more
- Tidyverse Packages for Importing
Cheatsheets for Posit and tidyverse packages

Tutorials

LearnR Interactive Tutorials:
- Wrangling penguins from Allison Horst - Covers dplyr only
- Interactive tutorials/exercises: from the Data Science in a Box course
Read:
- Working with Data from Data Carpentry, Ecology
- Tidying/reshaping tables using tidyr & joining data tables from Exploratory Data Analysis in R
Watch:
- Learning the R Tidyverse (~3.3 hrs) from LinkedIn Learning, Charlie Joey Hadley (Mason Log In)
- R data manipulation with RStudio and dplyr (~1hr) from UQ Library

Tidyverse Terms & Symbols

Most popular, our recommended choice. Written by Hadley Wickham (RStudio/Posit)
Uses words instead of notation and built to support piping.
- mydata |> filter(rows) |> select(columns)
- |> is now built into R, but the tidyverse originally used %>% (from magrittr)

dplyr is for manipulating one dataset
- filter() - Keep/Remove Rows/Observations
- select() - Keep/Remove Variables
- mutate() - Create or change variables
  - See the packages forcats, stringr, lubridate, hms and more for working with specific data types
- group_by() |> summarize() - Summarize/Collapse across rows
- also arrange(), relocate(), rename(), count(), and many more
tidyr is for combining datasets or restructuring a dataset
readr, haven, and many others are for importing data from other formats
purrr provides Tidyverse alternatives to the base R apply() series of functions