R: filter rows within timerange from previous rows

let’s imagine I have dataframe like this

I want to leave only rows with type ‘att’ that occurred within 15 sec after rows with type ‘def’.

So desired output is:

Could you help me please to figure it out?

>Solution :

You could use dplyr::filter() and dplyr::lag() here. Assuming your time variable is in datetime format:

library(dplyr)

df %>%
  filter(type == "att", 
         (time - dplyr::lag(time)) < 15, # could also use `(difftime(time, lag(time), units = "secs")) < 15`
         lag(type) == "def")

#   event                time type
# 1     2 2022-07-15 18:08:06  att
# 2     6 2022-07-15 18:13:36  att

Data:

df <- read.table(text = "event  time    type
1   '2022-07-15 18:08:05'   def
2   '2022-07-15 18:08:06'   att
3   '2022-07-15 18:09:00'   def
4   '2022-07-15 18:09:12'   def
5   '2022-07-15 18:13:26'   def
6   '2022-07-15 18:13:36'   att
7   '2022-07-15 18:19:05'   def
8   '2022-07-15 18:21:43'   def
9   '2022-07-15 18:26:06'   att
10  '2022-07-15 18:27:26'   def", h = TRUE)

df$time <- lubridate::as_datetime(df$time)