Pipe Operator

The pipe operator allows you to chain together a set of functions to conduct a sequence of manipulations on it. Conceptually, here’s what code written using the pipe operator looks like:

data |>
  step1 |>
  step2 |>
  step3

We start with our data. Then we do step1. Then we do step2. Then we do step3. The pipe ties it all together, enabling us to do multiple things to our data, all in one execution of code.

Pipe Operator

There are different approaches to writing code that performs multiple functions on the same object.

Here is the standard, non-pipe operator way:

# three steps: filter, calculate a mean, then select only some columns to keep
data_new <- filter(data, y != "c")
data_new <- mutate(data_new, x_mean = mean(x))
data_new <- select(data_new, y, x_mean)

An alternative is to use the pipe operator |>

# three steps: filter, calculate a mean, then select only some columns to keep
data_new <- data |>
  filter(y != "c") |>
  mutate(x_mean = mean(x)) |>
  select(y, x_mean)

Pipe Operator

# three steps: filter, calculate a mean, then select only some columns to keep
data_new <- data |>
  filter(y != "c") |>
  mutate(x_mean = mean(x)) |>
  select(y, x_mean)

With the pipe operator, the result of the previous line gets passed (or piped) onto the next function.

• The first line in this example is simply specifying the data frame that is being passed from one line to the next.

• Notice how I did not have to specify data inside the filter(), mutate(), and select(), functions. This makes the code more concise and easier to read.

• The end result of the last function, then gets assigned to data_new <-.

File Paths

R needs to know the full file path to the file on your computer in order to import it

On Macs:

Users/username/projects/project_name/a_file.csv

On Windows:

C:\username\projects\project_name\a_file.csv

Three approaches to getting the right file path

• Never… ever… ever… use option 1, setwd().

• Instead, you should use RProjects and here(). But we will not cover this until Class 5.

• For now, we can just use the RStudio Import Dataset GUI to find the absolute file path.

RStudio Import Dataset GUI

RStudio Import Dataset GUI

Types of Data Files

• csv

• tab-delimited

csv Data Files

tab-delimited Data Files

Importing Data Files

readr

library(readr)

# csv
data_import <- read_csv("filepath/datafile.csv")

# tab-delimited
data_import <- read_delim("filepath/datafile.txt", delim = "\t", 
                          escape_double = FALSE, trim_ws = TRUE)

• Use the RStudio Import Dataset GUI to get the filepath that you will need

• In Class 5 you will learn a better way to specify file paths using RProjects and here::here()

Merging Data Frames

You might find yourself in a situation where you need to import multiple data files and merge them into a single data frame. There are two general classes of merging data

Bind

Combining data frames together by stacking either the rows or columns

Bind

Combining data frames together by stacking either the rows or columns

Column Bind: same rows but different columns and stacks them side-by-side. This is a much less common situation than a row bind and can usually be accomplished with a join instead

Bind

dplyr

library(dplyr)

# row bind
data_merged <- bind_rows(data_1, data_2)

# column bind
data_merged <- bind_cols(data_1, data_2)

Join

Merging data frames together that have at least one column in common with a mix of shared and unique entries in that column (e.g. Subject IDs).

Full Join

For the most part you can get away with just knowing how to do a full join using full_join() from the dplyr package

data_merged <- full_join(data_1, data_2, by = "Subject")

data_merged <- full_join(data_1, data_2, by = c("Subject", "Session"))

Restructure

And actually, you can have a mix of wide and long formatted data in a single data frame.

Restructure

• A good rule of thumb for formatting data is to have your variables (IVs and DVs) each have their own column.

• This often results in:

  • Measured variables in wide format

  • Experimental conditions or repeated-measures in long format

• Restructuring data involves changing the structure from long-to-wide (wider) or wide-to-long (longer)

• The tidyr package provides useful functions to do this:

  • pivot_wider() “widens” data from long-to-wide

  • pivot_longer() “lengthens” data from wide-to-long

pivot_wider()

data_long <- data.frame(
  ParticipantID = rep(1:3, each = 3),
  TestType = rep(c("Stress Level", "Creativity", "Memory"), times = 3),
  Score = c(5, 7, 8, 3, 6, 7, 4, 8, 6)
  )

View(data_long)

pivot_wider()

The two main arguments to specify in pivot_wider() are

library(tidyr)

data_wide <- data_long |>
  pivot_wider(names_from = TestType,
              values_from = Score)

View(data_wide)

pivot_longer()

The three main arguments to specify in pivot_longer() are:

data_long_again <- data_wide |>
  pivot_longer(cols = any_of(c("Stress Level", "Creativity", "Memory")),
               names_to = "TestType",
               values_to = "Score")

View(data_long_again)

Mixed Wide and Long

Converting from a wide data format to a mix of wide and long can be more complicated

data_sessions_wide <- tibble(
  ParticipantID = 1:3,
  StressLevel_S1 = c(5, 6, 4),
  Creativity_S1 = c(7, 6, 8),
  Memory_S1 = c(8, 7, 6),
  StressLevel_S2 = c(4, 5, 3),
  Creativity_S2 = c(8, 7, 9),
  Memory_S2 = c(9, 8, 7),
  StressLevel_S3 = c(3, 4, 2),
  Creativity_S3 = c(9, 8, 10),
  Memory_S3 = c(10, 9, 8)
)

View(data_sessions_wide)

Mixed Wide and Long

Let’s restructure this to 4 columns: Session, Stress Level, Creativity, and Memory.

There are two strategies for doing this:

1. Using pivot_longer(), separate(), and then pivot_wider()
2. Use more complicated syntax in pivot_longer()

# 1. Using pivot_longer(), separate(), and then pivot_wider()
data_sessions_1 <- data_sessions_wide |>
  pivot_longer(cols = contains("_S"),
               names_to = "Session",
               values_to = "Score") |>
  separate(Session, into = c("Test", "Session")) |>
  pivot_wider(names_from = Test,
              values_from = Score)

# 2. Use more complicated syntax in pivot_longer()
data_sessions_2 <- data_sessions_wide |>
  pivot_longer(cols = contains("_S"),
               names_to = c(".value", "Session"),
               names_pattern = "(.*)_(S\\d)")

Tidy Select Functions