时间戳

首页 > 解决方案 > R：在 tibble 列中使用 strsplit

r - R：在 tibble 列中使用 strsplit

问题描述

我有一个小标题，其中一列是字符串。它们是调查受访者表示他们演奏的乐器的名称。我想捕获每个乐器，因为它是自己单独的字符串。此列中的值范围从像吉他这样的单弦到更复杂的答案：唱歌、鼓/打击乐、钢琴/键盘...等我尝试过这样的事情：

options <- strsplit(survey$instruments_list, "\\, | \\/ | ")

不幸的是，输出仍然在几个字符串之间有 / 字符。

同样作为最后一个问题，其中一位受访者的回答令人难以置信的冗长回答被多个空格隔开，我只想要乐器，而不是他们的生活故事。

任何建议将不胜感激，谢谢！编辑： dput(head(survey)) 的结果

structure(list(time_submitted = c("8/27/19 20:22", "8/29/19 12:15", 
"8/28/19 19:33", "8/29/19 16:25", "8/27/19 15:40", "8/27/19 22:59"
), pseudonym_generator = c("Fake rapper name generator", "Fake band name generator", 
"Fake band name generator", "Fake band name generator", "Fake band name generator", 
"Fake band name generator"), pseudonym = c("Lord Los Angeles", 
"Heroes War", "Puppets War", "West Magic", "Eller Angel", "Trace Stripes"
), sex = c("Male", "Male", "Male", "Male", "Male", "Male"), academic_major = c("Computer Science", 
"Computer Science", "Math", "Computer Science", "Computer Science", 
"Computer Science"), academic_level = c("Senior", "Junior", "Senior", 
"Junior", "Senior", "Senior"), year_born = c(1994, 1997, 1996, 
1999, 1998, 1986), instrument_list = c("Rap", "Guitar", "Guitar", 
"Trumpet", "Piano/Keyboards, Ukulele", NA), favorite_song_artist = c("40 crew", 
"Arctic Monkeys", "Avatar", "Ben Folds", "blink-182", "brian jonestown massacre / sarabeth tucek"
), favorite_song = c("Not Enough", "Arabella", "The Eagle Has Landed", 
"Still", "She's Out Of Her Mind", "Seer"), favorite_song_link = c("https://www.youtube.com/watch?v=uITuGZKljgQ", 
"https://www.youtube.com/watch?v=Jn6-TItCazo", "https://www.youtube.com/watch?v=4p6GWewmTYQ", 
"https://www.youtube.com/watch?v=ShBzUK4rnI8", "https://www.youtube.com/watch?v=krpm0v_486k", 
"https://youtu.be/C-XT7DZsNP8")), class = c("tbl_df", "tbl", 
"data.frame"), row.names = c(NA, -6L))

标签： rtibblestrsplit

解决方案

这个怎么样：

library(dplyr)
library(tidyr)
survey %>%
  transmute(pseudonym, inst = strsplit(instrument_list, "[,/]")) %>%
  filter(!is.na(inst)) %>%
  unnest() %>%
  mutate(inst = trimws(inst), plays = TRUE) %>%
  spread(inst, plays) %>%
  mutate_at(vars(-pseudonym), Negate(is.na))
# # A tibble: 5 x 7
#   pseudonym        Guitar Keyboards Piano Rap   Trumpet Ukulele
#   <chr>            <lgl>  <lgl>     <lgl> <lgl> <lgl>   <lgl>  
# 1 Eller Angel      FALSE  TRUE      TRUE  FALSE FALSE   TRUE   
# 2 Heroes War       TRUE   FALSE     FALSE FALSE FALSE   FALSE  
# 3 Lord Los Angeles FALSE  FALSE     FALSE TRUE  FALSE   FALSE  
# 4 Puppets War      TRUE   FALSE     FALSE FALSE FALSE   FALSE  
# 5 West Magic       FALSE  FALSE     FALSE FALSE TRUE    FALSE

将多种乐器组合成一个类别并不难。我将改编你的一种乐器来演示。

一种方法是 with case_when，也许是两种方法中更直接/更直接的一种：

survey %>%
  mutate(instrument_list = if_else(grepl("Lord", pseudonym), "Electric Guitar", instrument_list)) %>%
  transmute(pseudonym, inst = strsplit(instrument_list, "[,/]")) %>%
  filter(!is.na(inst)) %>%
  unnest() %>%
  mutate(inst = trimws(inst), plays = TRUE) %>%
  spread(inst, plays) %>%
  mutate_at(vars(-pseudonym), Negate(is.na))
# # A tibble: 5 x 7
#   pseudonym        `Electric Guitar` Guitar Keyboards Piano Trumpet Ukulele
#   <chr>            <lgl>             <lgl>  <lgl>     <lgl> <lgl>   <lgl>  
# 1 Eller Angel      FALSE             FALSE  TRUE      TRUE  FALSE   TRUE   
# 2 Heroes War       FALSE             TRUE   FALSE     FALSE FALSE   FALSE  
# 3 Lord Los Angeles TRUE              FALSE  FALSE     FALSE FALSE   FALSE  
# 4 Puppets War      FALSE             TRUE   FALSE     FALSE FALSE   FALSE  
# 5 West Magic       FALSE             FALSE  FALSE     FALSE TRUE    FALSE  

survey %>%
  mutate(instrument_list = if_else(grepl("Lord", pseudonym), "Electric Guitar", instrument_list)) %>%
  transmute(pseudonym, inst = strsplit(instrument_list, "[,/]")) %>%
  filter(!is.na(inst)) %>%
  unnest() %>%
  mutate(
    inst = case_when(
      grepl("\\bPiano\\b", inst, ignore.case = TRUE) ~ "Piano",
      grepl("\\bUkelete\\b", inst, ignore.case = TRUE) ~ "Ukelele",
      grepl("\\bGuitar\\b", inst, ignore.case = TRUE) ~ "Guitar",
      TRUE ~ trimws(inst)),
    plays = TRUE,
  ) %>%
  spread(inst, plays) %>%
  mutate_at(vars(-pseudonym), Negate(is.na))
# # A tibble: 5 x 6
#   pseudonym        Guitar Keyboards Piano Trumpet Ukulele
#   <chr>            <lgl>  <lgl>     <lgl> <lgl>   <lgl>  
# 1 Eller Angel      FALSE  TRUE      TRUE  FALSE   TRUE   
# 2 Heroes War       TRUE   FALSE     FALSE FALSE   FALSE  
# 3 Lord Los Angeles TRUE   FALSE     FALSE FALSE   FALSE  
# 4 Puppets War      TRUE   FALSE     FALSE FALSE   FALSE  
# 5 West Magic       FALSE  FALSE     FALSE TRUE    FALSE

另一种方法（如果你有更多）是合并/加入一个框架。这样做的一个优点是它可以非常具体并且包括非常不同的工具（其中正则表达式可能比您想要处理的更多）。一个缺点是它可能过于具体......例如，它不会加入拼写错误或大小写差异。

gen_inst <- tibble::tribble(
  ~inst, ~newinst
 ,"Electric Guitar", "Guitar"
 ,"Electric Bass"  , "Guitar"
 ,"Electric Piano" , "Piano"
 ,"Pipe Organ"     , "Piano"
)

survey %>%
  mutate(instrument_list = if_else(grepl("Lord", pseudonym), "Electric Guitar", instrument_list)) %>%
  transmute(pseudonym, inst = strsplit(instrument_list, "[,/]")) %>%
  filter(!is.na(inst)) %>%
  unnest() %>%
  left_join(gen_inst, by = "inst") %>%
  mutate(
    inst = if_else(is.na(newinst), trimws(inst), newinst),
    plays = TRUE
  ) %>%
  select(-newinst) %>%
  spread(inst, plays) %>%
  mutate_at(vars(-pseudonym), Negate(is.na))
# # A tibble: 5 x 6
#   pseudonym        Guitar Keyboards Piano Trumpet Ukulele
#   <chr>            <lgl>  <lgl>     <lgl> <lgl>   <lgl>  
# 1 Eller Angel      FALSE  TRUE      TRUE  FALSE   TRUE   
# 2 Heroes War       TRUE   FALSE     FALSE FALSE   FALSE  
# 3 Lord Los Angeles TRUE   FALSE     FALSE FALSE   FALSE  
# 4 Puppets War      TRUE   FALSE     FALSE FALSE   FALSE  
# 5 West Magic       FALSE  FALSE     FALSE TRUE    FALSE

推荐阅读