Iteration and simulation

Warm-up question

Problem: 10 people are at a party, and all of them are wearing hats. They each place their hat in a pile; when they leave, they choose a hat at random. What is the probability at least one person selected the correct hat?

Question: Work with your neighbor to discuss the following question:

• Without calculating probabilities, how could you design an experiment to estimate this probability?

Designing an experiment

Step 1: representing the hats

hats <- 1:10

hats

 [1]  1  2  3  4  5  6  7  8  9 10

hats[3]

[1] 3

• hats is a vector, containing the numbers 1 to 10
• entries in a vector are accessed by their index

Step 2: everyone draws a random hat

hats <- 1:10
randomized_hats <- sample(hats, size = 10, replace = FALSE)

hats

 [1]  1  2  3  4  5  6  7  8  9 10

randomized_hats

 [1]  4  7  1  3  8 10  9  2  5  6

• The sample function creates a random sample from a vector
• How many people selected their original hat?

Step 3: check who got their original hat

hats <- 1:10
randomized_hats <- sample(hats, size = 10, replace = FALSE)

hats

 [1]  1  2  3  4  5  6  7  8  9 10

randomized_hats

 [1]  4  7  1  3  8 10  9  2  5  6

hats == randomized_hats

 [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE

# TRUE is 1, FALSE is 0
sum(hats == randomized_hats)

[1] 0

# did at least one person get their hat?
sum(hats == randomized_hats) > 0

[1] FALSE

Code so far

hats <- 1:10
randomized_hats <- sample(hats, size = 10, replace = FALSE)

sum(hats == randomized_hats) > 0

[1] FALSE

• Is this a good estimate of the probability?

Step 4: iteration

A for loop repeats code many times:

nsim <- 10000 # number of simulations 
for(i in 1:nsim){
  
  
}

Step 4: iteration

A for loop repeats code many times:

nsim <- 10000 # number of simulations
hats <- 1:10
results <- rep(NA, nsim) # vector to store results

for(i in 1:nsim){
  randomized_hats <- sample(hats, size = 10, replace = FALSE)
  results[i] <- sum(hats == randomized_hats) > 0
}

head(results)

[1]  TRUE  TRUE FALSE  TRUE  TRUE  TRUE

Step 4: iteration

A for loop repeats code many times:

nsim <- 10000 # number of simulations
hats <- 1:10
results <- rep(NA, nsim) # vector to store results

for(i in 1:nsim){
  randomized_hats <- sample(hats, size = 10, replace = FALSE)
  results[i] <- sum(hats == randomized_hats) > 0
}

mean(results)

[1] 0.6231

• What if I wanted to repeat the simulation, with a different number of people?

Removing magic numbers

Without magic numbers:

nsim <- 10000 # number of simulations
n_people <- 10 # number of people
hats <- 1:n_people
results <- rep(NA, nsim) # vector to store results

for(i in 1:nsim){
  randomized_hats <- sample(hats, 
                            size = n_people, 
                            replace = FALSE)
  results[i] <- sum(hats == 
                      randomized_hats) > 0
}

mean(results)

[1] 0.6316

• Why did I get different results?

Final code

set.seed(3) # set a seed for reproducibility

n_people <- 10 # number of people at the party
hats <- 1:n_people # numbered hats
nsim <- 10000 # number of simulations
results <- rep(NA, nsim) # vector to store the results

for(i in 1:nsim){
  # hats are randomly assigned to each person
  randomized_hats <- sample(hats, n_people, replace = F)
  
  # did at least one person get their hat back?
  results[i] <- sum(randomized_hats == hats) > 0
}

mean(results)

Map?

set.seed(3) # set a seed for reproducibility

n_people <- 10 # number of people at the party
hats <- 1:n_people # numbered hats
nsim <- 10000 # number of simulations
results <- rep(NA, nsim) # vector to store the results

for(i in 1:nsim){
  # hats are randomly assigned to each person
  randomized_hats <- sample(hats, n_people, replace = F)
  
  # did at least one person get their hat back?
  results[i] <- sum(randomized_hats == hats) > 0
}

mean(results)

How could we do this with map instead?

Map: writing a function

set.seed(3)

hat_match <- function(n){
  hats <- 1:n
  randomized_hats <- sample(hats, n, replace = F)
  sum(randomized_hats == hats) > 0
}

hat_match(10)

[1] FALSE

hat_match(10)

[1] FALSE

hat_match(10)

[1] FALSE

hat_match(10)

[1] TRUE

Map: iterating

set.seed(3)
nsim <- 20

hat_match <- function(n){
  hats <- 1:n
  randomized_hats <- sample(hats, n, replace = F)
  sum(randomized_hats == hats) > 0
}

map(1:nsim, hat_match)

Will this do what I want?

Map: iterating

set.seed(3)
nsim <- 20
n_people <- 10

hat_match <- function(n){
  hats <- 1:n
  randomized_hats <- sample(hats, n, replace = F)
  sum(randomized_hats == hats) > 0
}

map_lgl(1:nsim, function(i) hat_match(n_people))

 [1] FALSE FALSE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE FALSE FALSE  TRUE  TRUE
[13]  TRUE  TRUE  TRUE FALSE  TRUE FALSE FALSE  TRUE

Map: iterating

set.seed(3)
nsim <- 1000
n_people <- 10

hat_match <- function(n){
  hats <- 1:n
  randomized_hats <- sample(hats, n, replace = F)
  sum(randomized_hats == hats) > 0
}

map_lgl(1:nsim, function(i) hat_match(n_people)) |>
  mean()

[1] 0.605

Class activity

https://sta279-f25.github.io/class_activities/ca_12.html

• Work with a neighbor on the class activity
• At the end of class, submit your work as an HTML file on Canvas (one per group, list all your names)