10  Heterogeneous Types

10.1 Lists

list, as.list, is.list, unlist

• Atomic vectors are collections of values of the same type.
• Lists are collections of vectors and lists. The vectors within this collection may be of any type, and of any length.
• Lists are recursive, meaning a list can be an element of a list.

image source

10.2 Creation

To create a list, we use the list function:

l1 <- list(    # l is a list of 3 elements
  1:3,            # 1. an integer vector of length 3
  "abc",          # 2. a character string vector of length 1
  c(1.2, 3.8)     # 3. a double vector of length 2
)
l1
typeof(l1)

A list can be an element of a list:

l2 <- list(    # l is a list of 2 elements
  list(           # 1. a list of 2 elements
    c(1, 2),           # 1.1 a double vector of length 2
    list("w")          # 1.2 a list of 1 element (which is a char vector of length 1)
  ),
  "a"             # 2. a character vector of length 1
)

The c function (implicitly) coerce vectors to lists of vectors before combining them into a list

l3 <- c(list(1, 2), c("3", "4"), list(5, 6))
l3

Illustration: list construction

Explicit coercion from other structures to lists is done via as.list.

v <- c(1, 2, 3)
# Are these two commands equivalent?
list(v)
as.list(v)

A list can be flatten-ed with unlist

v <- c(1, 2, 3)
# Are these two commands equivalent?
unlist(list(v))
unlist(as.list(v))

10.3 names attributes

Like in vectors, you can add attributes such as names to lists

l <- list(
  a = factor(c("apple", "apple", "banana")),
  b = c(1:10)
)
typeof(l[["a"]])
typeof(l)
names(l)
attributes(l)

One common usage of named lists is to return multiple values from a function:

# recall p1.3:
Bmi <- function(weight, height) {
  return(weight / height^2)
}

# now we want to know the value of bmi as well as if the bmi is showing a need to reduce weight:
Bmi2 <- function(weight, height) {
  bmi_value <- weight / height^2
  return(list(
    value = bmi_value,
    reduce_weight = bmi_value >= 25
  ))
}
bb <- Bmi2(weight = 90, height = 1.75)
bb[["value"]]
bb[["reduce_weight"]]

10.4 class attribute

The class attribute of a list determines how the list is printed, and which methods are used when we apply generic functions to the list. We will talk about generic functions and methods in [Object-oriented programming]. Examples of classes of lists include data.frame, tibble, lm, and glm:

df <- data.frame(x = 1:3, y = letters[1:3])
tb <- tibble::tibble(x = 1:3, y = letters[1:3])
lm1 <- lm(mpg ~ wt, data = mtcars)
glm1 <- glm(vs ~ wt, data = mtcars, family = binomial)
class(df)
class(tb)
class(lm1)
class(glm1)

10.5 data.frame and tibble

data.frame and tibble are two classes of the list data structure used in R:

• data.frame - Data frames are tightly coupled collections of variables. Data frame is a fundamental data structure used by most of R’s modelling functions.
• tibble - The data frames of tidyverse

10.5.1 data.frame

A data frame is a named list of vectors with attributes names, row.names, and class of “data.frame”. In contrast to regular lists, all vectors inside the collection of a data frame are required to have the same length. Thus a data frame is always rectangular shaped.

Because data frame shares the same shape with matrix, the same functions applicable to matrix also work on data frame: nrow, ncol, rownames, colnames, rbind, cbind, t

df <- data.frame(x = 1:3, y = letters[1:3])
df
attributes(df)

rownames(df)
colnames(df)
names(df)

nrow(df)   # same as matrix
ncol(df)
length(df) # same as ncol

t(df)

rbind(df, df)
cbind(df, df)

Noticed the column names are also duplicated? (See [Duplicated names])

To prove data frames are just lists with fancy attributes, we manually construct a data.frame object by adding attributes to a list:

l <- list(x = 1:3, y = letters[1:3])
attr(l, "class") <- "data.frame"
attr(l, "row.names") <- 1:3
identical(df, l)

Explicit coercion: as.data.frame

df1 <- as.data.frame(list(x = 1:3, y = letters[1:3]))
identical(df, df1)
is.data.frame(df1)

10.5.2 tibble

tibble is tidyverse’s version of data frame

library(tibble)
tb <- tibble(x = 1:3, y = letters[1:3])
typeof(tb)
attributes(tb)

is.data.frame(tb)