Btukfyl

I have this data:

library(tidyverse)

library(lubridate)



dates <- c("01/01/18 1:00:00 PM" ,"01/01/18 1:01:00 PM",

           "01/01/18 1:02:00 PM" ,"01/01/18 1:03:00 PM",

           "01/01/18 1:04:00 PM" ,"01/01/18 1:05:00 PM",

           "01/01/18 1:06:00 PM" ,"01/01/18 1:07:00 PM",

           "01/01/18 1:08:00 PM" ,"01/01/18 1:09:00 PM",

           "01/01/18 1:10:00 PM" ,"01/01/18 1:11:00 PM")



vals <- c(1, 2, 3, 3, 15, 16, 17, 18, 1, 2, 1, 22)



datfr <- data.frame(dates, vals)



datfr$dates <- dmy_hms(datfr$dates)

I want to apply a condition:

if the val is < 4 for 2 continuous minutes period then true

I tried :

datfr$gr <- datfr %>%

       group_by(by2min = cut(dates, "2 min")) %>%

       summarise(cond = (vals < 4))

but it gives me:

column cond must be length 1 not 2

and I am not sure for the approach.

So, my expected output:

dates                   vals     cond



1 2018-01-01 13:00:00     1      

2 2018-01-01 13:01:00     2      

3 2018-01-01 13:02:00     3      TRUE

4 2018-01-01 13:03:00     3     

5 2018-01-01 13:04:00    15      FALSE

6 2018-01-01 13:05:00    16     

7 2018-01-01 13:06:00    17      FALSE

8 2018-01-01 13:07:00    18      

9 2018-01-01 13:08:00     1      FALSE

10 2018-01-01 13:09:00    2      

11 2018-01-01 13:10:00    1      TRUE

12 2018-01-01 13:11:00   22      

Hence, if the for 2 continuous minutes the val is < 4 then true.

Let's assume your data is in a format where the time diffs are 1 minute between row entries

datfr$cond<-

zoo::rollapply(data = datfr$vals, width = 3, FUN = function(x) { if (all(x < 4)) return(TRUE) else return(FALSE) }, align = "right", fill = FALSE)

result:

#                 dates vals  cond

#1  2018-01-01 13:00:00    1 FALSE

#2  2018-01-01 13:01:00    2 FALSE

#3  2018-01-01 13:02:00    3  TRUE

#4  2018-01-01 13:03:00    3  TRUE

#5  2018-01-01 13:04:00   15 FALSE

#6  2018-01-01 13:05:00   16 FALSE

#7  2018-01-01 13:06:00   17 FALSE

#8  2018-01-01 13:07:00   18 FALSE

#9  2018-01-01 13:08:00    1 FALSE

#10 2018-01-01 13:09:00    2 FALSE

#11 2018-01-01 13:10:00    1  TRUE

#12 2018-01-01 13:11:00   22 FALSE

I've tried just to reproduce your desired output as close as it possible. I assume that empty elements of cond are NAs. In case if cond is a character variable, and empty elements represent s it's easy to adjust the output by adding additional mutate(cond = coalesce(as.character(cond), "")). I failed with conversion of the last value into s/NA.

#library(tidyverse)



datfr %>%

  arrange(dates) %>%

  group_by(by2min = lag(cut(c(min(dates), dates), "2 min"))[-1]) %>%

  mutate(dates = max(dates)) %>%

  group_by(dates) %>%

  summarise(cond = all(vals < 4), vals = last(vals)) %>%

  right_join(datfr, by = c('dates', 'vals')) %>%

  select(dates, vals, cond)



# # A tibble: 12 x 3

#   dates                vals cond 

#   <dttm>              <dbl> <lgl>

# 1 2018-01-01 13:00:00     1 NA   

# 2 2018-01-01 13:01:00     2 NA   

# 3 2018-01-01 13:02:00     3 TRUE 

# 4 2018-01-01 13:03:00     3 NA   

# 5 2018-01-01 13:04:00    15 FALSE

# 6 2018-01-01 13:05:00    16 NA   

# 7 2018-01-01 13:06:00    17 FALSE

# 8 2018-01-01 13:07:00    18 NA   

# 9 2018-01-01 13:08:00     1 FALSE

#10 2018-01-01 13:09:00     2 NA   

#11 2018-01-01 13:10:00     1 TRUE 

#12 2018-01-01 13:11:00    22 FALSE 

How about using rollapply?

zoo::rollapply(datfr$vals, 3, by = 1, function(x) sum(x<4) == 2)

Edit: simplified function