cm009 Exercises: tidy data

library(tidyverse)
lotr  <- read_csv("https://raw.githubusercontent.com/jennybc/lotr-tidy/master/data/lotr_tidy.csv")
guest <- read_csv("https://raw.githubusercontent.com/STAT545-UBC/Classroom/master/data/wedding/attend.csv")
email <- read_csv("https://raw.githubusercontent.com/STAT545-UBC/Classroom/master/data/wedding/emails.csv")
library(DT)

Exercise 1: Univariate Pivoting

Consider the Lord of the Rings data:

datatable(lotr)

1. Would you say this data is in tidy format?

• Quite tidy: Each observation is a combination of film, race, and gender. And each observation is how many words spoken.

2. Widen the data so that we see the words spoken by each race, by putting race as its own column.

# pivot_wider/pivot_longer equivalent gather/spread

(lotr_wide <- lotr %>% 
  pivot_wider(id_cols = c(-Race, -Words), 
              names_from = Race, 
              values_from = Words))

3. Re-lengthen the wide LOTR data from Question 2 above.

lotr_wide %>% 
  pivot_longer(cols = c(-Film, -Gender), 
               names_to  = "Race", 
               values_to = "Words")

lotr_wide %>%
  pivot_longer(cols = -c(Film, Gender),
               names_to = "Race",
               values_to = "Words")

Exercise 2: Multivariate Pivoting

Congratulations, you’re getting married! In addition to the wedding, you’ve decided to hold two other events: a day-of brunch and a day-before round of golf. You’ve made a guestlist of attendance so far, along with food preference for the food events (wedding and brunch).

guest %>% 
  DT::datatable(rownames = FALSE)

1. Put “meal” and “attendance” as their own columns, with the events living in a new column.

(guest_long <- guest %>% 
  pivot_longer(cols      = c(-party, -name), 
               names_to  = c(".value", "event"),
               names_sep = "_"))

2. Use tidyr::separate() to split the name into two columns: “first” and “last”. Then, re-unite them with tidyr::unite().

guest_long %>% 
  separate(name, into = c("first_name", "last_name"), sep=" ")

guest_long %>% 
  separate(name, into = c("first_name", "last_name"), sep=" ") %>%
  unite(col = "name", c("first_name", "last_name"), sep = " ")

3. Which parties still have a “PENDING” status for all members and all events?

guest_long %>% 
  group_by(party) %>% 
  summarize(all_pending = all(attendance == "PENDING"))

# all returns true for that party if attendance == "PENDING" for each party member
# all returns false for that party if attendance != "PEnDING" for one member of a party 

4. Which parties still have a “PENDING” status for all members for the wedding?

guest %>% 
  group_by(party) %>% 
  summarize(pending_wedding = all(attendance_wedding == "PENDING"))

5. Put the data back to the way it was.

guest_long %>% 
  pivot_wider(id_cols     = c(party,name), 
              names_from  = event, 
              names_sep   = "_", 
              values_from = c(meal,attendance))

6. You also have a list of emails for each party, in this worksheet under the variable email. Change this so that each person gets their own row. Use tidyr::separate_rows()

Exercise 3: Making tibbles

1. Create a tibble that has the following columns:

• A label column with "Sample A" in its entries.
• 100 random observations drawn from the N(0,1) distribution in the column x
• y calculated as the x values + N(0,1) error.

n <- 100
tibble(label = "Sample A",
             x = rnorm(n),
             y = x + rnorm(n))

2. Generate a Gaussian sample of size 100 for each combination of the following means (mu) and standard deviations (sd).

3. Fix the experiment tibble below (originally defined in the documentation of the tidyr::expand() function) so that all three repeats are displayed for each person, and the measurements are kept. The code is given, but needs one adjustment. What is it?

(experiment <- tibble(
  name = rep(c("Alex", "Robert", "Sam"), c(3, 2, 1)),
  trt  = rep(c("a", "b", "a"), c(3, 2, 1)),
  rep = c(1, 2, 3, 1, 2, 1),
  measurement_1 = runif(6),
  measurement_2 = runif(6)
))

experiment %>% complete(name,trt,rep)

experiment %>% complete(nesting(name, trt), rep)

experiment %>% expand(name, trt, rep)

# expand gives all possible combinations of inputs, even if not present in the dataset
# complete (without nesting) gives all possible combos with NAs
# complete (with nesting), completes the reps based on the existing name, trt combo