R / dyplr：将两行转换为两列

Question

我在 R 中有一个数据框，如下所示：

Word    Base    Number  Type
-       -       -       -
shoe    shoe    4834    singular
shoes   shoe    49955   plural
toy     toy     75465   singular
toys    toy     23556   plural
key     key     39485   singular
keys    key     6546    plural
jazz    jazz    58765   plural

我想改变它，使它看起来像这样：

Word_Sg Word_Pl Base    Num_Singular    Num_Plural
--      --      --      --              --
shoe    shoes   shoe    4834            49955
toy     toys    toy     75465           23556
key     keys    key     39485           6546
NA      jazz    jazz    NA              58765

因此，与其有两行单数和复数的值，我想要两列，一列是单数，另一列是复数。

我已经使用dplyr::summarize尝试了一些东西，但到目前为止，没有任何成功。 这是我到目前为止想出的代码：

dataframe1 <- dataframe %>% 
      mutate(Num_Singular = case_when(Type == "singular" ~ Number)) %>%
      mutate(Num_Plural = case_when(Type == "plural" ~ Number)) %>%
      dplyr::select(Word, Base, Num_Singular, Num_Plural) %>%
      group_by(Base) %>%
      dplyr::summarize(Num_Singular = paste(na.omit(Num_Singular)),
                       Num_Plural = paste(na.omit(Num_Plural))

但是，它给了我这个错误：

Error in summarise_impl(.data, dots) : 
  Column `Num_Singular` must be length 1 (a summary value), not 2)

我认为问题可能在于有些行不一定有单数和复数，但只有其中之一（例如“爵士”）。 大多数行都有。

那么我怎样才能在 R 或 dplyr 中做到这一点呢？

Answer 1

如果你先看前几列：

select(dat, Base, Word, Type)[1:2,]
#   Base  Word     Type
# 1 shoe  shoe singular
# 2 shoe shoes   plural

从这里开始，考虑它只是将其扩展为单数/复数列，有效地从“高”到“宽”。 （如果Type有两个以上的类别会更明显。）

select(dat, Base, Word, Type) %>%
    spread(Type, Word) %>%
    rename(Word_Pl=plural, Word_Sg=singular)
#   Base Word_Pl Word_Sg
# 1 jazz    jazz    <NA>
# 2  key    keys     key
# 3 shoe   shoes    shoe
# 4  toy    toys     toy

您也可以轻松地为Number重复此操作。 从那里开始，只需根据键列Base合并/加入它们即可：

full_join(
  select(dat, Base, Word, Type) %>%
    spread(Type, Word) %>%
    rename(Word_Pl=plural, Word_Sg=singular),
  select(dat, Base, Number, Type) %>%
    spread(Type, Number) %>%
    rename(Num_Pl=plural, Num_Sg=singular),
  by = "Base"
)
#   Base Word_Pl Word_Sg Num_Pl Num_Sg
# 1 jazz    jazz    <NA>  58765     NA
# 2  key    keys     key   6546  39485
# 3 shoe   shoes    shoe  49955   4834
# 4  toy    toys     toy  23556  75465

---

耗材数据：

library(dplyr)
library(tidyr)
dat <- read.table(text='Word    Base    Number  Type
shoe    shoe    4834    singular
shoes   shoe    49955   plural
toy     toy     75465   singular
toys    toy     23556   plural
key     key     39485   singular
keys    key     6546    plural
jazz    jazz    58765   plural', header=TRUE, stringsAsFactors=FALSE)

Answer 2

tidyr的新pivot_wider()函数使这变得简单......

library(dplyr)
library(tidyr)

dat <- read.table(header = T, stringsAsFactors = F, text='
Word    Base    Number  Type
shoe    shoe    4834    singular
shoes   shoe    49955   plural
toy     toy     75465   singular
toys    toy     23556   plural
key     key     39485   singular
keys    key     6546    plural
jazz    jazz    58765   plural')

dat %>% 
  pivot_wider(id_cols = Base, names_from = Type, values_from = c(Word, Number))

# # A tibble: 4 x 5
#   Base  Word_singular Word_plural Number_singular Number_plural
#   <chr> <chr>         <chr>                 <int>         <int>
# 1 shoe  shoe          shoes                  4834         49955
# 2 toy   toy           toys                  75465         23556
# 3 key   key           keys                  39485          6546
# 4 jazz  NA            jazz                     NA         58765

Answer 3

核心思想是通过它的类型以及它是一个单词还是一个数字来识别每个数据点......然后很容易传播到你想要的格式。 （我不会费心重命名变量或专门对它们进行排序以匹配您的预期输出，因为这很容易做到，而不是这里的问题的一部分）

library(dplyr)
library(tidyr)

dat <- read.table(header = T, stringsAsFactors = F, text='
Word    Base    Number  Type
shoe    shoe    4834    singular
shoes   shoe    49955   plural
toy     toy     75465   singular
toys    toy     23556   plural
key     key     39485   singular
keys    key     6546    plural
jazz    jazz    58765   plural')

dat %>% 
  gather(variable, value, Word, Number) %>%
  unite(Type, variable, Type) %>%
  spread(Type, value, convert = T) %>% 
  as_tibble()

# # A tibble: 4 x 5
#   Base  Number_plural Number_singular Word_plural Word_singular
#   <chr>         <int>           <int> <chr>       <chr>        
# 1 jazz          58765              NA jazz        NA           
# 2 key            6546           39485 keys        key          
# 3 shoe          49955            4834 shoes       shoe         
# 4 toy           23556           75465 toys        toy  

Answer 4

您可以通过Base加入数据的plural和singular子集，然后删除Type列并重新排序其他列......

full_join(filter(dat, Type == "plural"), 
          filter(dat, Type == "singular"),
          by = "Base", 
          suffix = c("_Pl", "_Sg")) %>% 
  select(Word_Sg, Word_Pl, Base, Number_Sg, Number_Pl)

#   Word_Sg Word_Pl Base Number_Sg Number_Pl
# 1    shoe   shoes shoe      4834     49955
# 2     toy    toys  toy     75465     23556
# 3     key    keys  key     39485      6546
# 4    <NA>    jazz jazz        NA     58765