当存在平局时如何总结数据集中前 3 个最高值

我有一个数据框 (my_data)，并且只想计算 3 个最高值的总和，即使可能存在平局。我对 R 很陌生并且我已经使用过dplyr.

A tibble: 15 x 3
   city      month number
   <chr>     <chr>  <dbl>
 1 Lund      jan       12
 2 Lund      feb       12
 3 Lund      mar       18
 4 Lund      apr       28
 5 Lund      may       28
 6 Stockholm jan       15
 7 Stockholm feb       15
 8 Stockholm mar       30
 9 Stockholm apr       30
10 Stockholm may       10
11 Uppsala   jan       22
12 Uppsala   feb       30
13 Uppsala   mar       40
14 Uppsala   apr       60
15 Uppsala   may       30

这是我尝试过的代码：

# For each city, count the top 3 of variable number
my_data %>% group_by(city) %>% top_n(3, number) %>% summarise(top_nr = sum(number))

预期（想要的）输出是：

# A tibble: 3 x 2
  city      top_nr
  <chr>      <dbl>
1 Lund          86
2 Stockholm     75
3 Uppsala      130

但实际的 R 输出是：

# A tibble: 3 x 2
  city      top_nr
  <chr>      <dbl>
1 Lund          86
2 Stockholm     90
3 Uppsala      160

似乎如果存在联系，则所有联系值都包含在求和中。我只想计算 3 个具有最高值的唯一实例。

任何帮助将非常感激！ :)

我们可以做一个distinct删除重复的元素。其中的方式top_n工作原理是，如果值重复，它将保留许多重复的行

my_data %>% 
   distinct(city, number, .keep_all = TRUE) %>%
   group_by(city) %>%
   top_n(3, number) %>%
   summarise(top_nr = sum(number))

Update

根据OP的新输出，之后top_n输出（这不是arranged)、将‘数字’按降序排列，得到sum前 3 个“数字”

my_data %>% 
   group_by(city) %>% 
   top_n(3, number) %>% 
   arrange(city,  desc(number)) %>% 
   summarise(number = sum(head(number, 3)))
# A tibble: 3 x 2
#  city      number
#  <chr>      <int>
#1 Lund          74
#2 Stockholm     75
#3 Uppsala      130

data

my_data <- structure(list(city = c("Lund", "Lund", "Lund", "Lund", "Lund", 
"Stockholm", "Stockholm", "Stockholm", "Stockholm", "Stockholm", 
"Uppsala", "Uppsala", "Uppsala", "Uppsala", "Uppsala"), month = c("jan", 
"feb", "mar", "apr", "may", "jan", "feb", "mar", "apr", "may", 
"jan", "feb", "mar", "apr", "may"), number = c(12L, 12L, 18L, 
28L, 28L, 15L, 15L, 30L, 30L, 10L, 22L, 30L, 40L, 60L, 30L)), 
class = "data.frame", row.names = c("1", 
"2", "3", "4", "5", "6", "7", "8", "9", "10", "11", "12", "13", 
"14", "15"))