根据条件由列元素变异

Question

我正在尝试改变一个列：

1. 数据框按错误分组
2. 变异列 (treatmentToGrowthRatio) 等于：

价值/（治疗==“成长”的价值）。

我有什么：

df <- tibble(
  type = as.factor(c("bug1", "bug1", "bug1", "bug2", "bug2", "bug2", "bug3", "bug3", "bug3", "blank")),
  treatment = c(rep(c("TreatA", "TreatB", "Growth"),3), "Blank"),
  value = 1:10
)

我正在执行的操作：

df %>% group_by(bug) %>% 
  mutate(
    treatmentToGrowthRatio =
      value/
      ## value where treatment == growth 
      ## (i.e. for bug 1 = 3; for bug 2 = 6; for bug 3 = 9; for Blank = NA)
  )

提供所需的输出：

dfFinal <- tibble(
  type = as.factor(c("bug1", "bug1", "bug1", "bug2", "bug2", "bug2", "bug3", "bug3", "bug3", "blank")),
  treatment = c(rep(c("TreatA", "TreatB", "Growth"),3), "Blank"),
  value = 1:10,
  treatmentToGrowthRatio = c(1/3, 2/3, 1, 4/6, 5/6, 1, 7/9, 8/9, 1, NA)
)

我得到的最接近的是treatmentToGrowthRatio = 1 where Treatment == "Growth" from:

df %>% group_by(type) %>% 
  mutate(
    treatmentToGrowthRatio =
      value/
      case_when(
        str_detect(treatment,
                   "Growth") ~ value
      )
  )

欣赏任何见解！谢谢。

Answer 1

您的问题陈述不清楚。组 type == "blank" 没有 treatment == "Growth"。在这种情况下，您预计会发生什么？

注意到我上面的评论，我将忽略带有type == "blank" 的行。然后你要么做

library(dplyr)
df %>%
    filter(type != "blank") %>%
    group_by(type) %>%
    mutate(treatmentToGrowthRatio = value / value[treatment == "Growth"]) %>%
    ungroup()
## A tibble: 9 x 4
#  type  treatment value treatmentToGrowthRatio
#  <fct> <chr>     <int>                  <dbl>
#1 bug1  TreatA        1                  0.333
#2 bug1  TreatB        2                  0.667
#3 bug1  Growth        3                  1    
#4 bug2  TreatA        4                  0.667
#5 bug2  TreatB        5                  0.833
#6 bug2  Growth        6                  1    
#7 bug3  TreatA        7                  0.778
#8 bug3  TreatB        8                  0.889
#9 bug3  Growth        9                  1     

或者（也许更优雅）从长到宽重新整形，然后从相关列中划分值。

library(dplyr)
library(tidyr)
df %>%
    pivot_wider(names_from = treatment) %>%
    mutate(across(starts_with("Treat"), ~ .x / Growth))
## A tibble: 4 x 5
#  type  TreatA TreatB Growth Blank
#  <fct>  <dbl>  <dbl>  <int> <int>
#1 bug1   0.333  0.667      3    NA
#2 bug2   0.667  0.833      6    NA
#3 bug3   0.778  0.889      9    NA
#4 blank NA     NA         NA    10

如有必要，然后再次整形。