Merge date and time, find the time difference in R - r

Question

I have a data set and there are some date and hour attributes. Here is the sample, then I will explain what I want to do;
date1
hour1
date2
hour2
date3
hour3
2014-03-16 00:00:00
16:20:00
2014-03-16 00:00:00
20:20:03
2014-03-16 00:00:00
22:12:34
2014-04-22 00:00:00
10:20:00
2014-04-22 00:00:00
15:20:03
2014-04-22 00:00:00
20:12:34
2015-03-12 00:00:00
16:20:00
2015-03-12 00:00:00
20:20:03
2015-03-12 00:00:00
22:12:34
We know event1 happens before event2 (event1 -> event2 -> event3)
But as you see, in the date attributes, time section is not correct yet we have hour attributes for each. What I want to do; I want to correct them by using hour attributes then find the difference between these two dates and create new attributes that gives the time difference as hours.
Sample for above table;
event2_time
4
5
4
I tried to merge hour to date and create a new attribute like this but it doesn't work.(my goal is actually correct the date value and get rid of the hour attribute)
trainTable <- trainTable %>%
mutate("newParam" = as.POSIXct(paste(alert_date, alert_hour), format="%Y-%m-%d %H:%M:%S")
I could use some help, thanks in advance.
Data
structure(list(alert_date = structure(c(1394928000, 1395014400,
1395014400, 1395187200, 1395273600, 1395014400), tzone = "UTC", class = c("POSIXct",
"POSIXt")), alert_hour = c("16:15:00", "20:53:00", "12:55:00",
"14:22:00", "12:07:00", "17:48:00"), firstInterv_date = structure(c(1394928000,
1395014400, 1395014400, 1395187200, 1395273600, 1395014400), tzone = "UTC", class = c("POSIXct",
"POSIXt")), firstInterv_hour = c("16:35:00", "21:05:00", "13:10:00",
"14:42:00", "12:07:00", "18:08:00"), extinction_date = structure(c(1394928000,
1395014400, 1395014400, 1395187200, 1395273600, 1395014400), tzone = "UTC", class = c("POSIXct",
"POSIXt")), extinction_hour = c("17:47:00", "22:46:00", "15:30:00",
"15:25:00", "13:14:00", "21:10:00")), row.names = c(NA, -6L), class = c("tbl_df",
"tbl", "data.frame"))

Answer 1

Try this solution using mapply. It uses strsplit to split the date from the hours.
dat <- as.data.frame( dat ) # tibbles are cool but sometimes very restrictive, so changing to data.frame here
dat_new <- data.frame( setNames( mapply( function(x,y){
tmp <- sapply( strsplit( as.character(dat[,x]), " "), function(z) z[1] );
list( as.POSIXct( paste(tmp,dat[,y] ) ) ) },
grep("date", colnames(dat)), grep("hour", colnames(dat)) ),
c("a","b","c") ) )
dat_new$b - dat_new$a
Time differences in secs
[1] 1200 720 900 1200 0 1200
# if you need tibbles convert back if you need
as_tibble( dat_new )
# A tibble: 6 x 3
a b c
<dttm> <dttm> <dttm>
1 2014-03-16 16:15:00 2014-03-16 16:35:00 2014-03-16 17:47:00
2 2014-03-17 20:53:00 2014-03-17 21:05:00 2014-03-17 22:46:00
3 2014-03-17 12:55:00 2014-03-17 13:10:00 2014-03-17 15:30:00
4 2014-03-19 14:22:00 2014-03-19 14:42:00 2014-03-19 15:25:00
5 2014-03-20 12:07:00 2014-03-20 12:07:00 2014-03-20 13:14:00
6 2014-03-17 17:48:00 2014-03-17 18:08:00 2014-03-17 21:10:00
Data
dat <- structure(list(alert_date = structure(c(1394928000, 1395014400,
1395014400, 1395187200, 1395273600, 1395014400), tzone = "UTC", class = c("POSIXct",
"POSIXt")), alert_hour = c("16:15:00", "20:53:00", "12:55:00",
"14:22:00", "12:07:00", "17:48:00"), firstInterv_date = structure(c(1394928000,
1395014400, 1395014400, 1395187200, 1395273600, 1395014400), tzone = "UTC", class = c("POSIXct",
"POSIXt")), firstInterv_hour = c("16:35:00", "21:05:00", "13:10:00",
"14:42:00", "12:07:00", "18:08:00"), extinction_date = structure(c(1394928000,
1395014400, 1395014400, 1395187200, 1395273600, 1395014400), tzone = "UTC", class = c("POSIXct",
"POSIXt")), extinction_hour = c("17:47:00", "22:46:00", "15:30:00",
"15:25:00", "13:14:00", "21:10:00")), row.names = c(NA, -6L), class = c("tbl_df",
"tbl", "data.frame"))