Row-wise operations in a dataframe where each row depends on previous row in R-CodePudding

I would like to find the combination of some value in a dataframe from one column and some row with the value from the same column and the previous row. Here is a reproducible example.

I have four possible conditions and I want to identify pairs of (two) successive rows.

library(tidyverse)
library(purrr)

df <- data.frame(
    condition = sample(
        x = c("IL", "IR", "EL", "ER"), size = 10, replace = TRUE
        )
    )

df

   condition
1         IR
2         EL
3         IR
4         EL
5         ER
6         IL
7         IR
8         EL
9         EL
10        EL

In other words, I would like to get a "pair" column such as:

   condition     pair
1         IR    NA
2         EL    IR-EL
3         IR    EL-IR
4         EL    IR-EL
5         ER    EL-ER
6         IL    ER-IL
7         IR    IL-IR
8         EL    IR-EL
9         EL    EL-EL
10        EL    EL-EL

I have tried using purrr::accumulate() in combination with paste() but it accumulates all previous values. How can I only keep the previous and current values (i.e., lag 1)?

df %>%
    mutate(
        pair = accumulate(
            .x = condition,
            .f = ~paste(.x, .y, sep = "-")
            )
        )

   condition                          pair
1         IL                            IL
2         ER                         IL-ER
3         IL                      IL-ER-IL
4         IL                   IL-ER-IL-IL
5         EL                IL-ER-IL-IL-EL
6         EL             IL-ER-IL-IL-EL-EL
7         IL          IL-ER-IL-IL-EL-EL-IL
8         IL       IL-ER-IL-IL-EL-EL-IL-IL
9         IL    IL-ER-IL-IL-EL-EL-IL-IL-IL
10        IR IL-ER-IL-IL-EL-EL-IL-IL-IL-IR

I only want to keep pairs of two successive trials. Any idea? NB: I would prefer a tidyverse option.

CodePudding user response：

Use lag.

df %>% 
  mutate(pair = paste(lag(condition),condition,sep="-"))

   condition  pair
1         ER NA-ER
2         EL ER-EL
3         IR EL-IR
4         ER IR-ER
5         ER ER-ER
6         IR ER-IR
7         IR IR-IR
8         IR IR-IR
9         IR IR-IR
10        EL IR-EL