r-根据时间分配变量

Question

我有一个时间序列，并且我有添加新变量的特定要求。

这是一些数据

dput(df)
structure(list(Time = structure(c(1567423339.229, 1567423399.018, 
1567424218.867, 1567425478.666, 1567425498.883, 1567426519.008, 
1567429378.848, 1567429398.979, 1567429978.723, 1567431218.909
), tzone = "", class = c("POSIXct", "POSIXt")), RaceNum = c("1", 
"1", "1", "1", "1", "1", "2", "2", "2", "2")), class = "data.frame", row.names = c(NA, 
-10L))

使用case_when或if else，我尝试在1:nrow的基础上分配d，除非下一个时间序列事件在1分钟内，否则它会使用上一个变量并添加一个b。正如你所看到的，当RaceNum发生变化时，编号会重新开始。一旦我建立了d，我就用RaceNum和cbind将df分开。

以下是的预期结果

dput(df2)
structure(list(Time = structure(c(1567423339.229, 1567423399.018, 
1567424218.867, 1567425478.666, 1567425498.883, 1567426519.008, 
1567429378.848, 1567429398.979, 1567429978.723, 1567431218.909
), tzone = "", class = c("POSIXct", "POSIXt")), RaceNum = c("1", 
"1", "1", "1", "1", "1", "2", "2", "2", "2"), d = c("1", "1b", 
"2", "3", "3b", "4", "1", "1b", "2", "3")), class = "data.frame", row.names = c(NA, 
-10L))

Answer 1

• 对于每个RaceNum，创建一个变量，当连续记录之间的差异大于1分钟时，该变量会递增
• 对于每组(d(，将letters粘贴到组号中

library(dplyr)
df %>%
group_by(RaceNum) %>%
mutate(d = cumsum(difftime(Time, lag(Time, default = first(Time)),
units = 'min') > 1) + 1) %>%
group_by(d, .add = TRUE) %>%
mutate(d = paste0(d, letters[row_number()]), 
#For 1st row remove a from 1a, 2a etc. 
d = ifelse(row_number() == 1, sub('a', '', d), d))  %>%
ungroup
#   Time                RaceNum d    
#   <dttm>              <chr>   <chr>
# 1 2019-09-02 19:22:19 1       1    
# 2 2019-09-02 19:23:19 1       1b   
# 3 2019-09-02 19:36:58 1       2    
# 4 2019-09-02 19:57:58 1       3    
# 5 2019-09-02 19:58:18 1       3b   
# 6 2019-09-02 20:15:19 1       4    
# 7 2019-09-02 21:02:58 2       1    
# 8 2019-09-02 21:03:18 2       1b   
# 9 2019-09-02 21:12:58 2       2    
#10 2019-09-02 21:33:38 2       3