使用值序列对行进行子集化并标识序列开始的列

Question

我有一个这种格式的大型数据集。我想 a) 识别具有以下值序列 1 - 1 - >1 - >1 x1 和 x10 之间任意位置的 ID/行；和 b) 生成一个新变量（“事件”），用于标识序列的开头，取值 X1,...,X10。

my_df <- data.frame(ID = c("a","b","c","d","e","f","g","h"),
                    replicate(8,sample(1:4,8,rep=TRUE)))

对于 a)，我将 >1 的值替换为 2，然后将值从 X1 粘贴到 X10，然后过滤序列 1 - 1 - 2 - 2。对于 b)，我使用嵌套创建了变量“event” ifelse() 来识别序列的开始位置。仅适用于 8 列。 有没有办法提高具有更多列的数据集的效率？

我非常感谢任何指点！

df_seq <- my_df%>%
  mutate_at(vars(starts_with('X')), funs(ifelse(. > 1, 2, .)))%>%
  mutate(seq = paste(X1,"-",X2,"-",X3,"-",X4,"-",X5,"-",X6,"-",X7,"-",X8))%>%
  filter(grepl("1 - 1 - 2 - 2", seq))%>%
  mutate(event = ifelse(X1 == 1 & X2 == 1 & X3 == 2 & X4 == 2,"X1",
                 ifelse(X2 == 1 & X3 == 1 & X4 == 2 & X5 == 2,"X2",
                 ifelse(X3 == 1 & X4 == 1 & X5 == 2 & X6 == 2,"X3",
                 ifelse(X4 == 1 & X5 == 1 & X6 == 2 & X7 == 2,"X4","X5")))))

Answer 1

你可以这样做：

library(dplyr)
library(tidyr)
library(stringr)

df |>
  unite(seq, starts_with("X"), sep = "", remove = FALSE) |>
  mutate(event = paste0("X", sapply(str_locate_all(seq, pattern ='11[2-9][2-9]'), function(x) x[, 1]['start'])),
         event = case_match(event, "XNA" ~ NA, .default = event)) 

输出示例：

  ID      seq X1 X2 X3 X4 X5 X6 X7 X8 event
1  a 13341322  1  3  3  4  1  3  2  2  <NA>
2  b 44141221  4  4  1  4  1  2  2  1  <NA>
3  c 14144333  1  4  1  4  4  3  3  3  <NA>
4  d 22414321  2  2  4  1  4  3  2  1  <NA>
5  e 21241341  2  1  2  4  1  3  4  1  <NA>
6  f 13411443  1  3  4  1  1  4  4  3    X4
7  g 31433142  3  1  4  3  3  1  4  2  <NA>
8  h 32114343  3  2  1  1  4  3  4  3    X3

使用数据：

> dput(df)
structure(list(ID = c("a", "b", "c", "d", "e", "f", "g", "h"), 
    X1 = c(1L, 4L, 1L, 2L, 2L, 1L, 3L, 3L), X2 = c(3L, 4L, 4L, 
    2L, 1L, 3L, 1L, 2L), X3 = c(3L, 1L, 1L, 4L, 2L, 4L, 4L, 1L
    ), X4 = c(4L, 4L, 4L, 1L, 4L, 1L, 3L, 1L), X5 = c(1L, 1L, 
    4L, 4L, 1L, 1L, 3L, 4L), X6 = c(3L, 2L, 3L, 3L, 3L, 4L, 1L, 
    3L), X7 = c(2L, 2L, 3L, 2L, 4L, 4L, 4L, 4L), X8 = c(2L, 1L, 
    3L, 1L, 1L, 3L, 2L, 3L)), class = "data.frame", row.names = c(NA, 
-8L))