如何在dplyr中按指定规则计算子组排名（GroupRank）？

需求说明 #

• 输入包含Element和Group的向量，其中Group≠0表示该Element需要划分为子组，且需与相邻同Group值的元素配对
• ElementCnt为每个Element在向量中出现的序号（按元素分组后的行号）
• 需要通过dplyr::mutate生成GroupRank列，规则如下：
  • 当Group=0时，GroupRank等于该行的ElementCnt
  • 当为子组（Group≠0）的首个元素时，GroupRank = 该元素的ElementCnt - 之前已出现的子组配对数
  • 当与上一行元素的Group值相同时，共享同一GroupRank

示例表格 #

原错误代码 #

以下代码无法正确计算GroupRank：

library(dplyr)

myData <-
  data.frame(
    Element = c("A","A","B","B","B","B","B","B","C","A","A","A","B","B","B"),
    Group = c(0,0,0,11,11,0,0,0,0,0,2,2,-3,-3,-3)
  )

excelCopy <- myData %>% 
  group_by(Element) %>% 
  mutate(ElementCnt = row_number()) %>% 
  mutate(GroupRank = case_when(Group > 0 ~ match(Group, unique(Group)),TRUE ~ ElementCnt)) %>%
  ungroup() 

print.data.frame(excelCopy)

修正后的解决方案代码 #

以下是适配测试场景的修正代码，可正确计算GroupRank：

excelCopy <- myData %>%
  mutate(origOrder = row_number()) %>%  # 记录原始行顺序，便于后续恢复
  group_by(Element) %>% 
  mutate(
    ElementCnt = row_number()
  ) %>%
  ungroup() %>%  
  mutate(
    Group = factor(Group, unique(Group))
  ) %>% 
  arrange(Group) %>% 
  mutate(
    groupCt = cumsum(Group != lag(Group, 1, Group[[1]])) - 1L
  ) %>%  
  group_by(Group) %>%  
  mutate(
    GroupRank = ElementCnt - groupCt,
    GroupRank = max(1,if_else( as.character(Group) == "0", ElementCnt, min(GroupRank))) # 增加最小值1，适配更多场景
  ) %>%  
  ungroup() %>% 
  arrange(origOrder) %>%  
  select(-groupCt)

内容的提问来源于stack exchange，提问作者Curious Jorge - user9788072