1

I wish to convert 10-minute time blocks into 1-minute time blocks for a large vector. My datatime values are spaced at 10-minutes intervals, which represent a 10-minute time block ending at the given time stamp. Example data below:

data<- structure(list(datetime = structure(1:7,  .Label = c("30/11/2011 02:32",
 "30/11/2011 02:42", "30/11/2011 02:52", "30/11/2011 03:02", "30/11/2011 03:12",
 "30/11/2011 03:22", "30/11/2011 03:32"), class = "factor"), count =
c(100L, 60L, 10L, 10L, 200L, 180L, 190L)), .Names = c("datetime", "count"), class =  
"data.frame", row.names = c(NA, -7L))

I have successfully converted the 10-minute time blocks into 1-minute blocks, between the first and last datetime values in my vector, using the xts package as follows: 

data$datetime <- strptime(data$datetime, format="%d/%m/%Y %H:%M") 
data$datetime <- as.POSIXct(data$datetime)
data_xts <- as.xts(data, order.by=data$datetime)
dtmins <- seq(start(data_xts$datetime), end(data_xts$datetime), by = "1 min")

However, as the time values relate to the ‘end’ of a 10-minute block, I also require nine rows of 1-minute datetime values to be created, prior to the first datetime row in my vector. I wish to end up with an output vector that looks like the below (and continues on to the last datetime value in my original vector):

datetime
30/11/2011 02:23
30/11/2011 02:24
30/11/2011 02:25
30/11/2011 02:26
30/11/2011 02:27
30/11/2011 02:28
30/11/2011 02:29
30/11/2011 02:30
30/11/2011 02:31
30/11/2011 02:32
30/11/2011 02:33
30/11/2011 02:34
30/11/2011 02:35
30/11/2011 02:36
30/11/2011 02:37
etc...

I am struggling to find a solution. Any help would be much appreciated!

Emily
  • 859
  • 5
  • 14
  • 31
  • did you look at `to.period`? – andreister Jul 01 '13 at 15:15
  • thanks! yes I did have a look, but this appears to only covert an object to a time period lower than itself, and does not give me a solution to creating the additional minute values that I wish. – Emily Jul 01 '13 at 16:12

1 Answers1

2

Just merge your xts object with the sequence of times you want.

Data <- structure(list(datetime=structure(c(1322641920, 1322642520, 1322643120,
  1322643720, 1322644320, 1322644920, 1322645520), class=c("POSIXct", "POSIXt"),
  tzone=""), count=c(100L, 60L, 10L, 10L, 200L, 180L, 190L)),
 .Names=c("datetime", "count"), row.names=c(NA, -7L), class="data.frame")

data_xts <- as.xts(Data[,-1], order.by=Data$datetime)
data_xts <- merge(data_xts, seq(start(data_xts)-60*9, end(data_xts), by="1 min"))

Then you can use na.locf or similar to fill in the missing values, if you want.

data_xts <- na.locf(data_xts)
Joshua Ulrich
  • 173,410
  • 32
  • 338
  • 418
  • Many thanks for your response, but you code does the same thing as mine (although it is much cleaner!). It only gives me an output list with minute values spanning from the start and end datetimes of the original list. I want to create an additional nine 1-minute datetime values, which start 9 minutes prior to the first datetime in my vector (as shown in the example). Thanks! – Emily Jul 01 '13 at 16:07
  • @Emily: then subtract 9 minutes from the start time: `start(data_xts)-60*9`. – Joshua Ulrich Jul 01 '13 at 16:09