Occasional tidyverse tasks

These tasks are “occasional”, until they’re not, and then you’ll be glad to have them.

library(tidyverse)
library(ggdist)

stringr

Lots of good string operations

library(stringr)

Squishing

a <- "   hello    "
print(a)

[1] "   hello    "

b <- str_squish(a)
print(b)

[1] "hello"

Detecting

x <- c("Sunday", "Monday", "March", "Thursday")

str_detect(x, "day")

[1]  TRUE  TRUE FALSE  TRUE

str_extract(x, "day")

[1] "day" "day" NA    "day"

Combining

first_name <- c("John", "Paul", "George", "Ringo")
last_name <- c("Lennon", "McCartney", "Harrison", "Starr")

band <- str_c(first_name, last_name, sep = " ")
print(band)

[1] "John Lennon"     "Paul McCartney"  "George Harrison" "Ringo Starr"    

str_flatten(band, collapse = ", ")

[1] "John Lennon, Paul McCartney, George Harrison, Ringo Starr"

“Glue”ing

str_glue(
  "I know {first_name} is in the band."
)

I know John is in the band.
I know Paul is in the band.
I know George is in the band.
I know Ringo is in the band.

forcats

Useful for doing things to factors

library(palmerpenguins)
library(phonTools)

data("pb52")

penguins |> 
  count(
    species
  ) ->
  species_count

Columns in annoying orders

species_count |> 
  ggplot(
    aes(species, n)
  )+
    geom_col()

pb52 |> 
  ggplot(
    aes(
      vowel, 
      f1,
      color = vowel,
      fill = vowel
    )
  )+
    stat_dots(
      side = "both",
      layout = "hex"
    )

Reordering with fct_reorder()

species_count |> 
  mutate(
    species = fct_reorder(
      species,
      desc(n)
    )
  ) |> 
  ggplot(
    aes(species, n)
  )+
    geom_col()

pb52 |> 
  mutate(
    vowel = fct_reorder(
      vowel,
      desc(f1),
      .fun = mean
    )
  ) |> 
  ggplot(
    aes(
      vowel,
      f1, 
      color = vowel,
      fill = vowel
    )
  )+
    stat_dots(
      side = "both",
      layout = "hex"
    )

lubridate

library(lubridate)

Dates are, in general, a mess. For example:

# this will change
# every time the site rebuilds
time_now <- now()
time_now

[1] "2024-09-20 15:20:18 UTC"

That looks like a reasonable time for us to read with our eyes, but if we convert it to a numeric value:

as.numeric(time_now)

[1] 1726845618

That number is how many seconds it has been since January 1, 1970 because that’s just the most stable way to represent dates and times.

Creating and parsing dates

# iso
ymd("2024-09-16")

[1] "2024-09-16"

# US
mdy("September 16, 2024")

[1] "2024-09-16"

# most of the world
dmy("16 September 2024")

[1] "2024-09-16"

dplyr

consecutive_id()

If you have data with “streaks” in it, consecutive_id() will return an ID vector for each streak.

commits <- c(
  "Oakley",
  "Skyler", "Skyler",
  "Robin",
  "Oakley", "Oakley"
)

consecutive_id(commits)

[1] 1 2 2 3 4 4

I’ve used this to re-create transcript-looking things from force-aligned data.

transcript <- read_csv("data/transcript.csv")
head(transcript)

transcript |> 
  filter(
    str_detect(
      tier,
      "words"
    )
  ) ->
  words

words |> 
  dplyr::slice(
    .by = tier,
    -1
  ) |>
  arrange(tmin) |> 
  drop_na() |> 
  mutate(
    turn  = consecutive_id(tier),
    speaker = str_remove(tier, " - words")
  ) ->
  words_turn

head(words_turn)

library(gt)

words_turn |> 
  summarise(
    .by = turn,
    speaker = first(speaker),
    start = min(tmin),
    end = max(tmax),
    phrase = str_flatten(text, collapse = " ")
  ) |> 
  gt() |> 
  fmt_number(
    columns = c(start, end),
    decimals = 3
  ) ->
  transcript_tbl

turn	speaker	start	end	phrase
1	IVR	0.672	10.832	well uh mister scott i have a number of uh things i'd like to ask you about i wonder if you'd just mind uh answering questions uh one
2	KY25A	10.702	11.002	yeah
3	IVR	10.832	13.352	after another if you if
4	KY25A	13.312	13.652	well
5	IVR	13.352	13.952	i remind
6	KY25A	13.652	14.972	now you
7	IVR	13.952	15.002	you
8	KY25A	15.002	15.312	might
9	IVR	15.212	15.312	of
10	KY25A	15.452	15.942	start
11	IVR	15.522	15.812	uh
12	KY25A	15.942	20.642	that i was born in eighteen sixty seven
13	IVR	21.922	24.712	mhm and that makes you how old ninety
14	KY25A	24.352	24.732	nintey
15	IVR	24.712	25.162	three
16	KY25A	24.732	25.172	three