时间戳

首页 > 解决方案 > mutate 和 case_when 可变数量的变量时的替代代码

r - mutate 和 case_when 可变数量的变量时的替代代码

问题描述

在下面的例子中

df <- data.frame(
  Sample = c(1:18),
  Group = c("A","A","A","A","A","A","A","A","A","B","B","B","B","B","B","B","B","B"), 
  B1 = c(34.4, 34.2, 34.1, 32.2, 32.8, 32.9, 30.8, 31.2, 30.5, 32.4, 32.4, 32.1, 31.2, 31.8, 31.9, 30.8, 30.2, 30.5),
  S1 = c("sample", "sample", "sample", "std", "std", "std", "sample", "sample", "sample","sample", "sample", "sample", "std", "std", "std", "sample", "sample", "sample")
)

B1我计算perGroup和 when 的最大值S1 == "std"

for(Var in unique(df$Group)) {
  assign(paste("maxB_", Var, sep = ""), max(filter(df, S1 == "std" & Group == Var)$B1, na.rm=T))
}

我想df$Bdl根据这段代码在 df 中创建一个新列

df %>% mutate(Bdl = case_when(!S1 == "std" & Group == "A" & B1 > maxB_A ~ "Y",
                                     !S1 == "std" & Group == "B" & B1 > maxB_B ~ "Y")
df$Bdl <- ifelse(df$S1!="std" & is.na(df$Bdl), "N", df$Bdl)

得到以下df

> df
   Sample Group   B1     S1  Bdl
1       1     A 34.4 sample    Y
2       2     A 34.2 sample    Y
3       3     A 34.1 sample    Y
4       4     A 32.2    std <NA>
5       5     A 32.8    std <NA>
6       6     A 32.9    std <NA>
7       7     A 30.8 sample    N
8       8     A 31.2 sample    N
9       9     A 30.5 sample    N
10     10     B 32.4 sample    Y
11     11     B 32.4 sample    Y
12     12     B 32.1 sample    Y
13     13     B 31.2    std <NA>
14     14     B 31.8    std <NA>
15     15     B 31.9    std <NA>
16     16     B 30.8 sample    N
17     17     B 30.2 sample    N
18     18     B 30.5 sample    N
>

当我有超过 2 个组时，我想使用此代码，而不必在 mutate 部分中添加所有这些代码/条件。我想出了以下方法，但我不确定这是否是最好的

test <- df %>% filter(S1 %in% "std") %>% group_by(Group) %>% summarise(maxB=max(B1)) 
df2 <- left_join(df, test) %>% mutate(Bdl = case_when(B1 > maxB ~ "Y"))  
df2$Bdl <- ifelse(df2$S1!="std" & is.na(df2$Bdl), "N", df2$Bdl)

标签： rdplyrcase-when

解决方案

您不需要for循环或临时变量，您可以在以下位置处理它mutate：

df %>%
  group_by(Group) %>%
  mutate(Bdl = if_else(S1 == "std", NA, B1 > B1[S1 == "std"])) %>%
  ungroup()
# # A tibble: 18 x 5
#    Sample Group    B1 S1     Bdl  
#     <int> <chr> <dbl> <chr>  <lgl>
#  1      1 A      34.4 sample TRUE 
#  2      2 A      34.2 sample TRUE 
#  3      3 A      34.1 sample TRUE 
#  4      4 A      32.2 std    NA   
#  5      5 A      32.8 std    NA   
#  6      6 A      32.9 std    NA   
#  7      7 A      30.8 sample FALSE
#  8      8 A      31.2 sample FALSE
#  9      9 A      30.5 sample FALSE
# 10     10 B      32.4 sample TRUE 
# 11     11 B      32.4 sample TRUE 
# 12     12 B      32.1 sample TRUE 
# 13     13 B      31.2 std    NA   
# 14     14 B      31.8 std    NA   
# 15     15 B      31.9 std    NA   
# 16     16 B      30.8 sample FALSE
# 17     17 B      30.2 sample FALSE
# 18     18 B      30.5 sample FALSE

我通常推荐/更喜欢存储logical而不是"Y"/ "N"，但如果你需要这些字母，你可以很容易地调整它：

df %>%
  group_by(Group) %>%
  mutate(Bdl = case_when(
    S1 == "std"          ~ NA_character_,
    B1 > B1[S1 == "std"] ~ "Y",
    TRUE                 ~ "N")
  ) %>%
  ungroup()

推荐阅读