How to insert rows at variable positions in a dataframe - r

Question

My original data frame shows changes on one variable (act, measured in seconds) over approx a 2-week period for several individuals (identified by Ring). My problem is that this variable stretches over the change of date (i.e. at midnight) and I wanted to split it in two: from time[i] till just at midnight, and from midnight until time[i+1]. I have added a few more variables that I need for computing these two operations:
modify the ith row (only when date changes) so I can get the portion of act[i] before midnight
insert one extra row (only when date changes) and assign it the other portion of act[i].
For example:
ith row: 01-01-2000 23:55:00 act= 360 seconds
i+1th row: 02-01-2000 00:01:00 act= 30 seconds
i+2th row: 02-01-2000 00:01:30 act= 50 seconds
.
.
.
My goal is to get:
ith row: 01-01-2000 23:55:00 act= 300 seconds # modified row
i+1th row: 02-01-2000 00:00:00 act= 60 seconds # inserted row
i+2th row: 02-01-2000 00:01:00 act= 30 seconds # previously row i+1th
i+3th row: 02-01-2000 00:01:30 act= 30 seconds #previously row i+2th
.
.
.
Data associated to each individual (Ring) stretch over a different period of time, thereby resulting in date changes between individuals that shoudn't be taken into account.
Below, a selection of my ~ 90000-row dataframe (xact) that shows date changes within and between individuals (Ring) and next my code:
Ring time act wd date clock timepos timemn actmn jul
156 6106933 09/06/11 21:37:45 267 dry 09/06/11 21:37:45 2011-06-09 21:37:45 2011-06-10 8535 15134
157 6106933 09/06/11 21:42:12 3417 wet 09/06/11 21:42:12 2011-06-09 21:42:12 2011-06-10 8268 15134
158 6106933 09/06/11 22:39:09 51 dry 09/06/11 22:39:09 2011-06-09 22:39:09 2011-06-10 4851 15134
159 6106933 09/06/11 22:40:00 7317 wet 09/06/11 22:40:00 2011-06-09 22:40:00 2011-06-10 4800 15134
160 6106933 10/06/11 00:41:57 24 dry 10/06/11 00:41:57 2011-06-10 00:41:57 2011-06-11 83883 15135
529 6106933 11/06/11 22:41:57 3177 wet 11/06/11 22:41:57 2011-06-11 22:41:57 2011-06-12 4683 15136
530 6106933 11/06/11 23:34:54 6 dry 11/06/11 23:34:54 2011-06-11 23:34:54 2011-06-12 1506 15136
531 6106933 11/06/11 23:35:00 1779 wet 11/06/11 23:35:00 2011-06-11 23:35:00 2011-06-12 1500 15136
532 6106933 12/06/11 00:04:39 594 dry 12/06/11 00:04:39 2011-06-12 00:04:39 2011-06-13 86121 15137
533 6106933 12/06/11 00:14:33 18840 wet 12/06/11 00:14:33 2011-06-12 00:14:33 2011-06-13 85527 15137
7024 6134701 24/07/11 15:24:14 6 dry 24/07/11 15:24:14 2011-07-24 15:24:14 2011-07-25 30946 15179
7025 6134701 24/07/11 15:24:20 6 wet 24/07/11 15:24:20 2011-07-24 15:24:20 2011-07-25 30940 15179
7026 6134701 24/07/11 15:24:26 810 dry 24/07/11 15:24:26 2011-07-24 15:24:26 2011-07-25 30934 15179
R = unique(xact$Ring)
for ( m in R ) {
for ( i in 1:nrow(xact) ) {
if( xact$jul[i] < xact$jul[i+1] ) {
# modify row i (jul= Julian date)
xact[i] <- c( xact$Ring[i], xact$time[i], xact$actmn[i], xact$wd[i], xact$date[i], xact$clock[i], xact$timepos[i], xact$timemn[i], xact$actmn[i], xact$jul[i] )
# add new row between row i and row i+1
r <- i
newrow <- c( xact$Ring[i], xact$timemn[i], as.numeric(xact$timepos[i+1] - xact$timemn[i]), xact$wd[i], xact$date[i+1], xact$clock[i+1], xact$timemn[i], xact$timemn[i], xact$actmn[i], xact$jul[i+1] )
insertRow <- function( xact, newrow, r ) {
xact[seq( r+1, nrow(xact) + 1), ] <- xact[seq( r, nrow(xact) ), ]
xact[r,] <- newrow
xact
}
}
}
}
I tried to adapt an existing code Add new row to dataframe, at specific row-index, not appended? but produces this message:
I would appreciate any help.
Santi

Answer 1

Here is an example with made-up data:
#create data
DF <- data.frame(time=seq(from=strptime("2013-01-01 01:00","%Y-%m-%d %H:%M"),to=strptime("2013-01-03 01:00","%Y-%m-%d %H:%M"),by=3500))
DF$ring <- 1:2
DF <- DF[order(DF$ID),]
#apply per ring
library(plyr)
DF <- ddply(DF,.(ring),function(df){
#index of date changes
ind <- c(FALSE,diff(as.POSIXlt(df$time)$yday)==1)
add <- df[ind,]
add$time <- round(add$time,"days")
#you can simply rbind and order, no need for inserting
df <- rbind(df,add)
df <- df[order(df$time),]
#it's easier to calculate act here
df$act <- c(diff(as.numeric(df$time)),NA)
df})
time ring act
1 2013-01-01 01:00:00 1 7000
2 2013-01-01 02:56:40 1 7000
3 2013-01-01 04:53:20 1 7000
4 2013-01-01 06:50:00 1 7000
5 2013-01-01 08:46:40 1 7000
6 2013-01-01 10:43:20 1 7000
7 2013-01-01 12:40:00 1 7000
8 2013-01-01 14:36:40 1 7000
9 2013-01-01 16:33:20 1 7000
10 2013-01-01 18:30:00 1 7000
11 2013-01-01 20:26:40 1 7000
12 2013-01-01 22:23:20 1 5800
13 2013-01-02 00:00:00 1 1200
14 2013-01-02 00:20:00 1 7000
15 2013-01-02 02:16:40 1 7000
16 2013-01-02 04:13:20 1 7000
17 2013-01-02 06:10:00 1 7000
18 2013-01-02 08:06:40 1 7000
19 2013-01-02 10:03:20 1 7000
20 2013-01-02 12:00:00 1 7000
21 2013-01-02 13:56:40 1 7000
22 2013-01-02 15:53:20 1 7000
23 2013-01-02 17:50:00 1 7000
24 2013-01-02 19:46:40 1 7000
25 2013-01-02 21:43:20 1 7000
26 2013-01-02 23:40:00 1 NA
27 2013-01-01 01:58:20 2 7000
28 2013-01-01 03:55:00 2 7000
29 2013-01-01 05:51:40 2 7000
30 2013-01-01 07:48:20 2 7000
31 2013-01-01 09:45:00 2 7000
32 2013-01-01 11:41:40 2 7000
33 2013-01-01 13:38:20 2 7000
34 2013-01-01 15:35:00 2 7000
35 2013-01-01 17:31:40 2 7000
36 2013-01-01 19:28:20 2 7000
37 2013-01-01 21:25:00 2 7000
38 2013-01-01 23:21:40 2 2300
39 2013-01-02 00:00:00 2 4700
40 2013-01-02 01:18:20 2 7000
41 2013-01-02 03:15:00 2 7000
42 2013-01-02 05:11:40 2 7000
43 2013-01-02 07:08:20 2 7000
44 2013-01-02 09:05:00 2 7000
45 2013-01-02 11:01:40 2 7000
46 2013-01-02 12:58:20 2 7000
47 2013-01-02 14:55:00 2 7000
48 2013-01-02 16:51:40 2 7000
49 2013-01-02 18:48:20 2 7000
50 2013-01-02 20:45:00 2 7000
51 2013-01-02 22:41:40 2 4700
52 2013-01-03 00:00:00 2 2300
53 2013-01-03 00:38:20 2 NA