R 数据表，与不寻常的格式化表融为一体

我有一个类似以下的数据表：

data.table(`COLOUR` = c("red", "blue", "green"), 
           `COUNT(1:1)` = 5:7, 
           `COUNT(2:1)` = 1:3, 
           `COUNT(1:2)` = 1:3, 
           `COUNT(2:2)` = 6:8, 
           `RATE(1:1)` = 1/(1:3), 
           `RATE(2:1)` = 2/(2:4), 
           `RATE(1:2)` = 3/(4:6), 
           `RATE(2:2)` = 1,
           `INDICATOR(2)` = c("left", "lefter", "leftest")
           `INDICATOR(1)` = c("right", "righter", "more right"))

structure(list(COLOUR = c("red", "red", "red", "red", "blue", 
"blue", "blue", "blue", "green", "green", "green", "green"), 
    x = c(1, 1, 2, 2, 1, 1, 2, 2, 1, 1, 2, 2), y = c(1, 2, 1, 
    2, 1, 2, 1, 2, 1, 2, 1, 2), INDICATOR = c("right", "left", 
    "right", "left", "righter", "lefter", "righter", "lefter", 
    "more right", "leftest", "more right", "leftest"), COUNT = c(5, 
    2, 1, 6, 6, 2, 2, 7, 7, 3, 3, 8), RATE = c(1, 0.75, 1, 1, 
    0.5, 0.6667, 0.6, 1, 0.333, 0.5, 0.5, 1)), row.names = c(NA, 
-12L), class = c("data.table", "data.frame"), .internal.selfref = <pointer: 0x0000014d26c41ef0>)

我需要

melt

it以提供类似于以下内容的长期结果：

data.table(`COLOUR` = c("red", "blue", "green"), 
           `COUNT(1:1)` = 5:7, 
           `COUNT(2:1)` = 1:3, 
           `COUNT(1:2)` = 1:3, 
           `COUNT(2:2)` = 6:8, 
           `RATE(1:1)` = 1/(1:3), 
           `RATE(2:1)` = 2/(2:4), 
           `RATE(1:2)` = 3/(4:6), 
           `RATE(2:2)` = 1,
           `INDICATOR(2)` = c("left", "lefter", "leftest")
           `INDICATOR(1)` = c("right", "righter", "more right"))

structure(list(COLOUR = c("red", "red", "red", "red", "blue", 
"blue", "blue", "blue", "green", "green", "green", "green"), 
    x = c(1, 1, 2, 2, 1, 1, 2, 2, 1, 1, 2, 2), y = c(1, 2, 1, 
    2, 1, 2, 1, 2, 1, 2, 1, 2), INDICATOR = c("right", "left", 
    "right", "left", "righter", "lefter", "righter", "lefter", 
    "more right", "leftest", "more right", "leftest"), COUNT = c(5, 
    2, 1, 6, 6, 2, 2, 7, 7, 3, 3, 8), RATE = c(1, 0.75, 1, 1, 
    0.5, 0.6667, 0.6, 1, 0.333, 0.5, 0.5, 1)), row.names = c(NA, 
-12L), class = c("data.table", "data.frame"), .internal.selfref = <pointer: 0x0000014d26c41ef0>)

结构（列表（颜色=c（“红色”、“红色”、“红色”、“红色”、“红色”、“蓝色”）， “蓝色”、“蓝色”、“蓝色”、“绿色”、“绿色”、“绿色”、“绿色”）， x=c（1,1,2,2,1,1,2,2,1,1,2,2），y=c（1,2,1， 2,1,2,1,2,1,1,2,1,2），指示器=c（“右”、“左”， “右”，“左”，“右”，“左”，“右”，“左”， “更右”、“最左”、“更右”、“最左”），计数=c（5， 2,1,6,6,2,2,7,7,3,3,8），比率=c（1,0.75,1,1， 0.5,0.6667,0.6,1,0.333,0.5,0.5,1）），row.names=c（NA， -12L），class=c（“data.table”，“data.frame”），.internal.selfref=） 您会注意到，原始日期集中的列在其列名中存储了两个变量，分别是

COUNT

和

RATE

：x和

y

值，

指示符

列在其列名中存储

y

值

---

我希望使用

melt

来实现这一点，我怀疑它需要多个应用程序，但我不太确定最好的方法是什么。

我们可以分两步来实现这一点

library(data.table)
nm1 <- names(dt1)[!grepl("^INDICATOR", names(dt1))]
dt2 <- melt(dt1[, ..nm1], id.var = c("COLOUR"))[, 
         c('variable', 'x', 'y') := tstrsplit(variable, "[():]")][]       

dt3 <- melt(dt1[, .SD, .SDcols = patterns("^(COLOUR|INDICATOR)")], 
    id.var = 'COLOUR', value.name = 'INDICATOR')[,
          y := sub(".*\\((\\d+)\\)", "\\1", variable)][]
dcast(dt2[dt3, on = .(COLOUR, y)], COLOUR + INDICATOR + x + y ~ 
         variable, value.var = 'value')[
   order(factor(COLOUR, levels = c('red', 'blue', 'green')), x, y)]
#   COLOUR  INDICATOR x y COUNT      RATE
# 1:    red      right 1 1     5 1.0000000
# 2:    red       left 1 2     1 0.7500000
# 3:    red      right 2 1     1 1.0000000
# 4:    red       left 2 2     6 1.0000000
# 5:   blue    righter 1 1     6 0.5000000
# 6:   blue     lefter 1 2     2 0.6000000
# 7:   blue    righter 2 1     2 0.6666667
# 8:   blue     lefter 2 2     7 1.0000000
# 9:  green more right 1 1     7 0.3333333
#10:  green    leftest 1 2     3 0.5000000
#11:  green more right 2 1     3 0.5000000
#12:  green    leftest 2 2     8 1.0000000

库（data.table）
nm1我们可以将
pivot\u longer
与
names\u pattern
参数一起使用，这样我们在各自的列中就有了数据。我们可以根据
y
值选择
指标
列

library(dplyr)

tidyr::pivot_longer(dt, 
                cols = -c(COLOUR,starts_with('INDICATOR')),
                names_to = c('.value', 'x', 'y'), 
                names_pattern = '(.*)\\((\\d+):(\\d+)\\)') %>%
      mutate(INDICATOR = ifelse(y == 1, `INDICATOR(1)`, `INDICATOR(2)`)) %>%
      select(-c(`INDICATOR(1)`, `INDICATOR(2)`))


# A tibble: 12 x 6
#   COLOUR x     y     COUNT  RATE INDICATOR 
#   <chr>  <chr> <chr> <int> <dbl> <chr>     
# 1 red    1     1         5 1     right     
# 2 red    2     1         1 1     right     
# 3 red    1     2         1 0.75  left      
# 4 red    2     2         6 1     left      
# 5 blue   1     1         6 0.5   brighter   
# 6 blue   2     1         2 0.667 brighter   
# 7 blue   1     2         2 0.6   letter    
# 8 blue   2     2         7 1     letter    
# 9 green  1     1         7 0.333 more right
#10 green  2     1         3 0.5   more right
#11 green  1     2         3 0.5   leftist   
#12 green  2     2         8 1     leftest   

库（dplyr）
tidyr：：枢轴_更长（dt，
cols=-c（颜色，以（'INDICATOR'）开头），
名称_to=c（'.value'，x'，y'），
名称\u模式='（.*）\\（\\d+：（\\d+\\））%>%
突变（指标=ifelse（y==1，`INDICATOR（1）`，`INDICATOR（2）`））%>%
选择（-c（`INDICATOR（1）`，`INDICATOR（2）`））
#一个tibble:12x6
#彩色x y计数率指示器
#              
#1红色1 1 5 1右侧
#2红色2 1右侧
#3红色1 2 1 0.75左
#4红色2 2 6 1左
#5蓝1 16 0.5亮
#6蓝色2 1 2 0.667亮
#7个蓝色1 2 0.6字母
#8蓝色2 2 7 1字母
#9绿色1 17 0.333更多右侧
#10绿色21 30.5更多右侧
#11绿色1230.5左派
#12绿色2 2 8 1最左边