R：确定组内某些子组的最大值和最小值之间的差异

Question

下面的示例数据字段

Event        Ethnicity        Score
50 yd dash    Asian             7
50 yd dash    Afr. Am           8
50 yd dash    White             5
Hurdle        Asian             6
Hurdle        Afr. Am           8
Hurdle        White             9

我正在尝试确定每个事件中某些种族之间的差异，希望使用 dplyr 或 tidyverse 中的某些东西，但会采取任何 answer/help。比如亚组和白组在每场比赛中的差距，

例如，亚洲人 (7) - 白人 (5) = 差异 (2)，

导致类似于以下的输出：

Event          Difference
50 yd dash         2
Hurdle            -3

Answer 1

使用以下内容应该可以帮助您：

library(tidyverse)

df %>%
    spread(Ethnicity, Score) %>%
    mutate("Difference" = Asian - White) %>%
    select(-Asian, -White, -`Afr. Am`)
#       Event Difference
#1 50 yd dash          2
#2     Hurdle         -3

数据。

df <-
structure(list(Event = structure(c(1L, 1L, 1L, 2L, 2L, 2L), .Label = c("50 yd dash", 
"Hurdle"), class = "factor"), Ethnicity = structure(c(2L, 1L, 
3L, 2L, 1L, 3L), .Label = c("Afr. Am", "Asian", "White"), class = "factor"), 
    Score = c(7L, 8L, 5L, 6L, 8L, 9L)), class = "data.frame", row.names = c(NA, 
-6L))

@AntoniosK 已经发布了 read.table 读取 OP 发布的数据的方法，但我的方法有点不同。我没有从列的值中删除空格，而是将它们放在单引号之间。（它必须是单引号，因为指令将参数 text 的值放在双引号之间。）

df <- read.table(text = "
Event        Ethnicity        Score
'50 yd dash'    Asian             7
'50 yd dash'    'Afr. Am'           8
'50 yd dash'    White             5
Hurdle        Asian             6
Hurdle        'Afr. Am'           8
Hurdle        White             9
", header = TRUE)

Answer 2

df %>%
  mutate(rn = row_number()) %>%
  spread(Ethnicity, Score) %>%
  group_by(Event) %>%
  summarise(Difference = max(Asian, na.rm = T) - max(White, na.rm = T))

# # A tibble: 2 x 2
#   Event      Difference
#   <chr>           <dbl>
# 1 50 yd dash          2
# 2 Hurdle             -3

数据：

df <- 
structure(list(Event = c("50 yd dash", "50 yd dash", "50 yd dash", 
"Hurdle", "Hurdle", "Hurdle"), Ethnicity = c("Asian", "Afr. Am", 
"White", "Asian", "Afr. Am", "White"), Score = c(7, 8, 5, 6, 
8, 9)), row.names = c(NA, -6L), class = c("tbl_df", "tbl", "data.frame"
))

Answer 3

数据

df = read.table(text = "
Event        Ethnicity        Score
50yddash    Asian             7
50yddash    Afr.Am           8
50yddash    White             5
Hurdle        Asian             6
Hurdle        Afr.Am           8
Hurdle        White             9
", header=T, stringsAsFactors=F)

第一种方法，您手动指定感兴趣的种族：

library(dplyr)

df %>%
  group_by(Event) %>%
  summarise(Diff = Score[Ethnicity=="Asian"] - Score[Ethnicity=="White"])

# # A tibble: 2 x 2
#   Event     Diff
#   <chr>    <int>
# 1 50yddash     2
# 2 Hurdle      -3

您可以将这段代码用作函数（输入感兴趣的两个种族）。

第二种方法，您可以为所有独特的种族和事件组合创建所有差异：

library(tidyverse)

# create vectorised function that calculates the difference
# based on a given event and ethnicities
f = function(event, eth1, eth2) {
  df$Score[df$Event==event & df$Ethnicity==eth1] -
  df$Score[df$Event==event & df$Ethnicity==eth2] }
f = Vectorize(f)


data.frame(t(combn(unique(df$Ethnicity), 2)), stringsAsFactors = F) %>% # create combinations of ethnicities
  mutate(Event = list(unique(df$Event))) %>%                            # create combinations with events
  unnest() %>%
  mutate(Diff = f(Event, X1, X2))                                       # apply the function

#    X1     X2    Event Diff
# 1  Asian Afr.Am 50yddash   -1
# 2  Asian Afr.Am   Hurdle   -2
# 3  Asian  White 50yddash    2
# 4  Asian  White   Hurdle   -3
# 5 Afr.Am  White 50yddash    3
# 6 Afr.Am  White   Hurdle   -1

此过程使用字母顺序来创建独特的差异。如果你想要所有这些（即亚裔白人和白人亚洲人），你可以使用这个

expand.grid(Event = unique(df$Event),
            X1 = unique(df$Ethnicity),
            X2 = unique(df$Ethnicity)) %>%
  filter(X1 != X2) %>%
  mutate(Diff = f(Event, X1, X2))                                     

#    Event     X1     X2 Diff
# 1  50yddash Afr.Am  Asian    1
# 2    Hurdle Afr.Am  Asian    2
# 3  50yddash  White  Asian   -2
# 4    Hurdle  White  Asian    3
# 5  50yddash  Asian Afr.Am   -1
# 6    Hurdle  Asian Afr.Am   -2
# 7  50yddash  White Afr.Am   -3
# 8    Hurdle  White Afr.Am    1
# 9  50yddash  Asian  White    2
# 10   Hurdle  Asian  White   -3
# 11 50yddash Afr.Am  White    3
# 12   Hurdle Afr.Am  White   -1

R：确定组内某些子组的最大值和最小值之间的差异

R: determining difference between max and min for certain subgroups within groups

grouping

r

difference

dplyr

tidyverse