R：使用 dplyr 根据值变化添加新变量

Question

我有如下数据集：

id <- c(1,1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2,2)
v1 <- c("A","A","A","A","B","B","B","B","B","A","A","A","A","A","A","A","A","B","B","B","B","B","A","A","A","A")  
v2 <- c(1,1,1,2,2,2,2,2,2,2,3,3,3,1,1,1,2,2,2,2,2,2,2,3,3,3)
mydata<- data.frame(id,v1,v2)

v1表示移动与否，条件A=移动，B=不移动，v2是每个id经过的位置序列，1->2->3。我想添加一个列，说明是否：id 从 1 移动到 2，id 保持在 2，id 从 2 移动到 3。

具体来说，我尝试根据以下条件向数据框添加一列：if v1=A (id moving) and v2=1 or (v2=2 if previously v2 had been 1) phase= "去”，如果 v1=B（不动）阶段“停留”并且如果 v1=A（移动）和 v2=3 或（v2=2 在“停留”阶段之后）阶段 =“返回”。

具体来说，我的预期输出是：

phase <- c("Going","Going","Going","Going","Staying","Staying","Staying","Staying","Staying","Returning","Returning","Returning","Returning","Going","Going","Going","Going","Staying","Staying","Staying","Staying","Staying","Returning","Returning","Returning","Returning")

mydata2 <- cbind(mydata, phase)

我尝试了以下方法：

mydata <-mydata %>% group_by(id)%>% mutate(phase= case_when(v1 == "A" & (v2 == 1 | v2==2) ~ "Going", phase == "A" & (v2 == 2 | v2==3) ~ "Returning", phase == "B" ~ "Staying"))

但是 (v2 == 1 | v2==2) 不考虑值的顺序。有任何想法吗？如果听起来很复杂，我们深表歉意，很高兴进一步澄清。

Answer 1

按'id'分组后，根据相邻元素是否相同（rleid）在'v1'上创建一个分组，然后使用case_when，如果'v1' 值为“B”，return 'Staying'，如果 'v1' 为 'A' 且 'grp' 为 1，则 'Going' 且如果 'grp' 大于 1，则 'Returning'

library(data.table)
library(dplyr)
mydata %>% 
   group_by(id) %>%
   mutate(grp = rleid(v1), phase = case_when(v1 == 'B' ~ 'Staying',
        v1 == 'A' & grp == 1 ~ 'Going', v1 == 'A' &
          grp > 1 ~ 'Returning')) %>%
   ungroup %>%
   select(-grp)

R：使用 dplyr 根据值变化添加新变量

R: add new variable based on value change using dplyr

r

dplyr

case-when