Lists vs Data Frame

[Flippie.Coetser]

I can achieve the desired outcome using five different techniques.
The five basic examples below uses either a list or a data frame.

Since all these example work, my question is how performance and memory usage differs.
Meaning fundamentally, while desired functionality can be achieved with a list or data frame, which is a better option?

I am very interested to hear your thoughts.

# Utility function used by examples
printName <- \(item) print(item[["name"]])

# Data Frame Example 1
id <- "1234"
name <- "name"
student.one <- data.frame(id, name)

student.one |> printName()

# Data Frame Example 2
student.two <- data.frame(
  id = "1234",
  name = "name"
)

student.two |> printName()

# List Example 3
student.three <- NULL
student.three[["id"]]   <- "1234"
student.three[["name"]] <- "name"

student.three |> printName()

# List Example 4
student.four <- list(id="1234",name="name")

student.four |> printName()

# Mix Example 5
student.five <- data.frame(student.four)

student.five |> printName()

[nirgrahamuk]

Two is superior to One ( one makes additional objects outside of the final result that are probably a waste) - and to has less tying / repetition

Similarly Four is superior to 3 for the programmer as its less to type/ less to read

Five can be thought of as One with extra steps

[Flippie.Coetser]

Thanks @nirgrahamuk for the input. Your feedback makes sense. Apart from the flexibility lists offer, I wonder which data structure is more performant: lists or dataframes. Maybe I should try to set up some benchmarking and do many reads and writes to both and see what the results says

[RYann]

hey.
You can use the tictoc package and tic() expression toc()

I edited your code. thing is, your code is really simple, and i cant see actual differences. maybe overloading the system will produce more differenes between your methods.

#install.packages("tictoc")
library(tictoc)

# Utility function used by examples
printName <- \(item) print(item[["name"]])

# Data Frame Example 1
tic("example 1")
id <- "1234"
name <- "name"
student.one <- data.frame(id, name)

student.one |> printName()
toc()
tic("example 2")
# Data Frame Example 2
student.two <- data.frame(
  id = "1234",
  name = "name"
)

student.two |> printName()
toc()
tic("example 3")
# List Example 3
student.three <- NULL
student.three[["id"]]   <- "1234"
student.three[["name"]] <- "name"

student.three |> printName()
toc()
tic("example 4")
# List Example 4
student.four <- list(id="1234",name="name")

student.four |> printName()
toc()
tic("example 5")
# Mix Example 5
student.five <- data.frame(student.four)

student.five |> printName()
toc()

[Flippie.Coetser]

@RYann thanks for pointing me to the tictoc library. I did some tests for both memory usage and speed. Conclusion: List takes less memory and is faster. Here is a copy of the code I used to do the testing:

library(tictoc)

# Comparing the memory usage of a Person object create using different techniques
# Example arguments
id        <- 12345
firstname <- 'John'
lastname  <- 'Doe'
age       <- 30
country   <- 'Germany'
city      <- 'Berlin'

Person <- list()

# Using Data Frame
Person[["DataFrame.1"]] <- \(id, firstname, lastname, age, country, city) data.frame(
  Id        = id,
  Firstname = firstname,
  Lastname  = lastname,
  Age       = age,
  Country   = country,
  City      = city
)

Person[["DataFrame.2"]] <- \(id, firstname, lastname, age, country, city) {
  person <- data.frame(
    Id = id,
    Firstname = firstname,
    Lastname =  lastname,
    Age = age,
    Country = country,
    City = city
  )
}

Person[["DataFrame.3"]] <- \(id, firstname, lastname, age, country, city) {
  person <- data.frame(
    Id = id,
    Firstname = firstname,
    Lastname =  lastname,
    Age = age,
    Country = country,
    City = city
  )

  return(person)
}

# Using Lists
Person[["List.1"]] <- \(id, firstname, lastname, age, country, city) list(
  Id = id,
  Firstname = firstname,
  Lastname =  lastname,
  Age = age,
  Country = country,
  City = city
)

Person[["List.2"]] <- \(id, firstname, lastname, age, country, city) {
  person <- list(
    Id = id,
    Firstname = firstname,
    Lastname =  lastname,
    Age = age,
    Country = country,
    City = city
  )
}

Person[["List.3"]] <- \(id, firstname, lastname, age, country, city) {
  person <- list(
    Id = id,
    Firstname = firstname,
    Lastname =  lastname,
    Age = age,
    Country = country,
    City = city
  )

  return(person)
}


object.size(Person[["DataFrame.1"]]) # Results = 3064 bytes
object.size(Person[["DataFrame.3"]]) # Results = 7632 bytes
object.size(Person[["DataFrame.2"]]) # Results = 6568 bytes

object.size(Person[["List.1"]])      # Results = 3064 bytes
object.size(Person[["List.2"]])      # Results = 6568 bytes
object.size(Person[["List.3"]])      # Results = 7632 bytes

person.one <- data.frame(
  Id = id,
  Firstname = firstname,
  Lastname =  lastname,
  Age = age,
  Country = country,
  City = city
)
person.two   <- Person[["DataFrame.1"]](id, firstname, lastname, age, country, city)
person.three <- Person[["DataFrame.2"]](id, firstname, lastname, age, country, city)
person.four  <- Person[["DataFrame.3"]](id, firstname, lastname, age, country, city)

person.five  <- list(
  Id = id,
  Firstname = firstname,
  Lastname =  lastname,
  Age = age,
  Country = country,
  City = city
)
person.six   <- Person[["List.1"]](id, firstname, lastname, age, country, city)
person.seven <- Person[["List.2"]](id, firstname, lastname, age, country, city)
person.eight <- Person[["List.3"]](id, firstname, lastname, age, country, city)

# Memory Usage
# Data Frame based
object.size(person.one)   # Results 1616 bytes
object.size(person.two)   # Results 1616 bytes
object.size(person.three) # Results 1616 bytes
object.size(person.four)  # Results 1616 bytes

# List Based
object.size(person.five)  # Results 1216 bytes
object.size(person.six)   # Results 1216 bytes
object.size(person.seven) # Results 1216 bytes
object.size(person.eight) # Results 1216 bytes

PrintPersonInfo <- \(person) {
  print(person[["Id"]])
  print(person[["Firstname"]])
  print(person[["Lastname"]])
  print(person[["Age"]])
  print(person[["Country"]])
  print(person[["City"]])
}

CreatePerson <- \(createFunction,
                   printInfoFunction,
                   id, firstname, lastname, age, country, city) {
  tic()
  for (i in 1:100) {
    person <- createFunction(id, firstname, lastname, age, country, city)
    person |> printInfoFunction()
  }
  toc()
}

CreatePerson(Person[["DataFrame.1"]],PrintPersonInfo,id, firstname, lastname, age, country, city) # Results 0.11 sec
CreatePerson(Person[["DataFrame.2"]],PrintPersonInfo,id, firstname, lastname, age, country, city) # Results 0.11 sec
CreatePerson(Person[["DataFrame.3"]],PrintPersonInfo,id, firstname, lastname, age, country, city) # Results 0.11 sec
CreatePerson(Person[["List.1"]],PrintPersonInfo,id, firstname, lastname, age, country, city) # Results 0.05 sec
CreatePerson(Person[["List.2"]],PrintPersonInfo,id, firstname, lastname, age, country, city) # Results 0.02 sec
CreatePerson(Person[["List.3"]],PrintPersonInfo,id, firstname, lastname, age, country, city) # Results 0.04 sec