Subtract time of the day (H:M form) in columns - r

Question

Here is subset of my original data I am working with:
dput(datumi)
structure(c("21:26", "21:33", "21:38", "23:02", "23:03", "21:27",
"21:34", "21:39", "23:03", "23:04", "21:26", "21:33", "21:38",
"23:02", "23:04", "21:26", "21:34", "21:38", "23:02", "23:04",
"21:27", "21:34", "21:39", "23:02", "23:04"), .Dim = c(5L, 5L
), .Dimnames = list(c("2", "3", "4", "5", "6"), c("Datum_1",
"Datum_2", "Datum_3", "Datum_4", "Datum_5")))
So I am working with time, where e.g., 21:26 means time of the day.
Now I would like to subtract second column from first one and third from second and so on, this means that I would subtract column Datum_2 from Datum_1 and column Datum_3 from Datum_2 and Datum_4 from Datum_3. And my output will be new columns with differences in seconds
I've already created function/loop that does this if my data would be numeric, so for example in case of numeric data I would do this and get the desired output:
dat <- data.frame(
column1 = round(runif(n = 10, min=0, max=5),0),
column2 = round(runif(n = 10, min=0, max=5),0),
column3 = round(runif(n = 10, min=0, max=5),0),
column4 = round(runif(n = 10, min=0, max=5),0)
)
results <- list()
for(i in 1:length(dat)) {
if (i==length(dat)){
results[[i]] <-dat[,i]
} else {results[[i]] <-dat[,i+1] - dat[,i]}
}
results <- t(do.call(rbind,results))
results <- data.frame(results)
But I cannot figure it out for time format and I have tried strptime and as.POSIXct
x1 <- strptime(datumi, "%H:%M")
as.numeric(x1,units="secs")
and
as.POSIXct(datumi,format="%H:%M")
And also looked at this
Subtract time in r
Subtracting Two Columns Consisting of Both Date and Time in R
convert character to time in R

Answer 1

Here is one solution based on the answer given in R: Convert hours:minutes:seconds.
datumi
# Datum_1 Datum_2 Datum_3 Datum_4 Datum_5
# 2 "21:26" "21:27" "21:26" "21:26" "21:27"
# 3 "21:33" "21:34" "21:33" "21:34" "21:34"
# 4 "21:38" "21:39" "21:38" "21:38" "21:39"
# 5 "23:02" "23:03" "23:02" "23:02" "23:02"
# 6 "23:03" "23:04" "23:04" "23:04" "23:04"
makeTime <- function(x) as.POSIXct(paste(Sys.Date(), x))
dat <- apply(datumi, 2, makeTime)
mapply(x = 2:ncol(dat),
y = 1:(ncol(dat) -1),
function(x, y) dat[ , x] - dat[ , y])
# [,1] [,2] [,3] [,4]
# [1,] 60 -60 0 60
# [2,] 60 -60 60 0
# [3,] 60 -60 0 60
# [4,] 60 -60 0 0
# [5,] 60 0 0 0
You can also use as.POSIXct without pasting the current data with the 'format' argument:
makeTime <- function(x) as.POSIXct(x, format = "%H:%M")
Note, the result is the same because as.POSIXct assumes the current date when none is given.

Answer 2

One way you could also do it if you wanted to have column names in addition to your original data would be to do:
df<-as.data.frame(lapply(dat,strptime,format="%H:%M"))
lapply(1:4, function(i) df[,paste0("diff",i,"_",i+1)] <<- difftime(df[,i],df[,i+1],units=c("secs")))
df
Datum_1 Datum_2 Datum_3 Datum_4 Datum_5 diff1_2 diff2_3 diff3_4
2 2016-07-22 21:26:00 2016-07-22 21:27:00 2016-07-22 21:26:00 2016-07-22 21:26:00 2016-07-22 21:27:00 -60 secs 60 secs 0 secs
3 2016-07-22 21:33:00 2016-07-22 21:34:00 2016-07-22 21:33:00 2016-07-22 21:34:00 2016-07-22 21:34:00 -60 secs 60 secs -60 secs
4 2016-07-22 21:38:00 2016-07-22 21:39:00 2016-07-22 21:38:00 2016-07-22 21:38:00 2016-07-22 21:39:00 -60 secs 60 secs 0 secs
5 2016-07-22 23:02:00 2016-07-22 23:03:00 2016-07-22 23:02:00 2016-07-22 23:02:00 2016-07-22 23:02:00 -60 secs 60 secs 0 secs
6 2016-07-22 23:03:00 2016-07-22 23:04:00 2016-07-22 23:04:00 2016-07-22 23:04:00 2016-07-22 23:04:00 -60 secs 0 secs 0 secs
diff4_5
2 -60 secs
3 0 secs
4 -60 secs
5 0 secs
6 0 secs

Answer 3

I've found solution to my problem including function/loop that I've created for numeric data, I just needed to include
difftime(strptime(datumi[,i+1], format = "%H:%M"), strptime(datumi[,i], format = "%H:%M"), units = "secs") in my for loop function so code looks like this
datumi <- as.data.frame(datumi)
results <- list()
for(i in 1:length(dat)) {
if (i==length(dat)){
results[[i]] <-NULL
} else {results[[i]] <-difftime(strptime(datumi[,i+1], format = "%H:%M"), strptime(datumi[,1], format = "%H:%M"), units = "secs") }
}
results <- t(do.call(rbind,results))
results <- data.frame(results)
#And output
X1 X2 X3 X4
2 60 0 0 60
3 60 0 60 60
4 60 0 0 60
5 60 0 0 0
6 60 60 60 60
But because mapply used by #dayne is more convenient for me (because it applys function to multiple list arguments and is more readable for me) I used his solution.