play the game on the site, and stay or switch.
after a few rounds the game shows your rounds as R code, one line per fact.
a type case keeps many small things in order because every column holds only one kind of thing.
image: ai-generated (gpt-image-2.5-flare)
four shapes of data
in a matrix every cell has the same type, in a data frame every column has its own.
one game, four facts
image: ai-generated (gpt-image-2.5-flare)
play a few rounds
rounds as vectors
prize_door <- c(2, 1, 3, 3, 1)
contestant_choice <- c(2, 3, 1, 1, 1)
monty_opens <- c(1, 2, 2, 2, 3)
decision <- c("stay", "switch", "stay", "switch", "switch")
c() combines values into a vector, and the first value of each vector belongs to the first round.
the same position, the same round
four vectors of the same length describe five rounds, as long as nobody mixes up the order.
one vector, one type
every value in a vector has the same type, and typeof() tells you which one.
asking a vector
length(prize_door) # 5
typeof(decision) # "character"
mean(prize_door) # 2
table(decision) # stay 2, switch 3
sum, mean, median, min, max, range, sd, sort, unique, table: functions that take a whole vector at once.
when types meet
R never mixes types in a vector, it quietly converts to the most general one.
comparing whole vectors
an operation on two vectors works position by position, with no loop to write.
counting with TRUE and FALSE
first_pick_right <- prize_door == contestant_choice
sum(first_pick_right) # 2
mean(first_pick_right) # 0.4
R counts TRUE as 1 and FALSE as 0, so the mean of a logical vector is a share.
picking values by position
decision[1] # "stay"
decision[-1] # all but the first
decision[2:4] # "switch" "stay" "switch"
decision[c(1, 3)] # "stay" "stay"
square brackets select, a minus leaves out, and the result is again a vector.
picking values by a condition
a logical vector in square brackets keeps exactly the positions that are TRUE.
when a value is missing
NA means "not known", so anything computed with it is not known either, until you say what to do.
factors: text with fixed levels
decision_factor <- factor(
decision,
levels = c("stay", "switch")
)
levels(decision_factor) # "stay" "switch"
a factor knows all its possible values and their order, even those that do not occur.
a vector is a row of values of one type, and position i belongs to observation i.
vectors become a table
games <- data.frame(prize_door, contestant_choice, monty_opens, decision)
a data frame is a list of equally long vectors, and every vector holds values of one type.
tidy data
after Wickham, Çetinkaya-Rundel & Grolemund: R for Data Science, chapter "data tidying"
when a table is tidy, every column is a vector of one variable, and the tools of R fit it.
data frame and tibble
read.csv("lets_make_a_deal.csv")
86 Linda 2 1 3 stay
87 Robert 3 3 2 stay
88 Barbara 1 3 2 stay
89 Karen 2 2 1 stay
90 Sarah 1 3 2 switch
91 Joseph 1 1 3 stay
92 Richard 1 2 3 switch
93 Charles 1 1 3 stay
94 Susan 3 1 2 stay
95 Patricia 1 3 2 stay
96 Richard 1 2 3 stay
97 Patricia 1 3 2 switch
98 Susan 3 3 1 stay
99 Jessica 1 2 3 stay
100 Patricia 1 1 3 stay
read_csv("lets_make_a_deal.csv")
# A tibble: 100 × 5
contestant prize_door contestant_choice monty_opens decision
<chr> <dbl> <dbl> <dbl> <chr>
1 Charles 3 3 2 stay
2 Patricia 1 3 2 stay
3 Karen 3 3 1 stay
4 Jessica 3 1 2 stay
5 Joseph 1 3 2 stay
6 Robert 1 3 2 stay
7 John 3 1 2 stay
8 Richard 2 3 1 switch
9 Michael 3 1 2 stay
10 Elizabeth 3 2 1 switch
# ℹ 90 more rows
a tibble is a data frame with better manners: it shows the types and does not flood the console.
100 games as a tibble
library(tidyverse)
monty_hall <- read_csv("lets_make_a_deal.csv")
glimpse(monty_hall)
Rows: 100
Columns: 5
$ contestant <chr> "Charles", "Patricia", "Karen", "Jessica", "…
$ prize_door <dbl> 3, 1, 3, 3, 1, 1, 3, 2, 3, 3, 3, 3, 1, 3, 2,…
$ contestant_choice <dbl> 3, 3, 3, 1, 3, 3, 1, 3, 1, 2, 3, 2, 1, 3, 1,…
$ monty_opens <dbl> 2, 2, 1, 2, 2, 2, 2, 1, 2, 1, 2, 1, 2, 2, 3,…
$ decision <chr> "stay", "stay", "stay", "stay", "stay", "sta…
glimpse() turns the table on its side and shows every column with its type and first values; it is the first thing to run on new data.
a data frame is a list of equally long vectors, and every vector holds values of one type.
to read on: Hands-On Programming with R, chapter "R objects", and R for Data Science, chapters "data tidying", "logical vectors", "factors" and "missing values"