R dplyr按组执行不同的聚合

Posted 2021-04-30

我有一个数据帧dat，看起来像这样：

   dat <- structure(list(cell.ID = c(329574L, 329574L, 329574L, 329574L, 
    329574L, 329574L, 329574L, 329574L, 329574L, 329574L, 329574L, 
    329574L), Year = c("2010", "2010", "2010", "2010", "2010", "2010", 
    "2010", "2010", "2010", "2010", "2010", "2010"), month_name = c("June", 
    "July", "June", "July", "June", "July", "June", "July", "June", 
    "July", "June", "July"), value = c(459.860986624053, 398.94083733151, 
    16, 23, 111.69, 453.333, 71.55, 30.38, 31.928, 30.13355, 17.587, 
    19.7938709677419), variable_name = c("ETo", "ETo", "Rday", "Rday", 
    "Rsum", "Rsum", "Thdd", "Thdd", "Tmax", "Tmax", "Tmin", "Tmin"
    ), monthID = c(6L, 7L, 6L, 7L, 6L, 7L, 6L, 7L, 6L, 7L, 6L, 7L
    )), row.names = c(NA, -12L), class = "data.frame")


library(dplyr)

dat  %>%
dplyr::group_by(Year, variable_name) %>% 
dplyr::summarise(variable = sum(value))

如果我想对Tmax和Tmin求平均值，并对其余变量求和，则执行此操作

dat %>%        
dplyr::group_by(Year, variable_name) %>% 
dplyr::summarise(variable = ifelse(variable_name %in% c('Tmax', 'Tmin'), mean(value), sum(value)))

Error: Column `variable` must be length 1 (a summary value), not 2  

我该如何纠正？

答案

I think问题是，在这种情况下，ifelse是按行操作的，而不是在组的级别上。如果是正确的话，则可以通过获取两个摘要统计信息，然后有条件地通过变量名称选择想要的变量，来解决此问题，如下所示：

dat %>%        
dplyr::group_by(Year, variable_name) %>% 
dplyr::summarise(var_mean = mean(value), var_sum = sum(value)) %>%
dplyr::mutate(variable = ifelse(variable_name %in% c('Tmax', 'Tmin'), var_mean, var_sum)) %>%
dplyr::select(-var_mean, -var_sum)

结果：

# A tibble: 6 x 3
# Groups:   Year [1]
  Year  variable_name variable
  <chr> <chr>            <dbl>
1 2010  ETo              859. 
2 2010  Rday              39  
3 2010  Rsum             565. 
4 2010  Thdd             102. 
5 2010  Tmax              31.0
6 2010  Tmin              18.7

另一答案

[另一种方法是dplyr是使用if和else而不是ifelse：

dat %>%        
  group_by(Year, variable_name) %>% 
  summarise(variable = if (variable_name[1] %in% c('Tmax', 'Tmin')) mean(value) else sum(value))

# A tibble: 6 x 3
# Groups:   Year [1]
  Year  variable_name variable
  <chr> <chr>            <dbl>
1 2010  ETo              859. 
2 2010  Rday              39  
3 2010  Rsum             565. 
4 2010  Thdd             102. 
5 2010  Tmax              31.0
6 2010  Tmin              18.7