R - interpolate time series to same interval but new time points - r

Question

Apologies if this is an obvious question, but I am new to R, having spent many years with SAS
I have two data files of measurements taken on 10-second intervals, but not at the exact same time points. I would like to convert one of the time series to match the times of the other, adjusting its values with linear interpolation. In SAS I could do this pretty quickly with proc expand, but I can't find anything similar package in R (I've looked at zoo and xts).
To show what I mean, here are snippets of my two files. In this case one time series is on the 3's and the other is on the 2's. In this particular case that 1-second difference is probably trivial, but this is a problem I run into a lot in my work and so I'd like to know if there's any easy way to recalculate, via linear interpolation, the values in the second data set to also be on the 3's
Date,Time,Value
3-Nov-16,13:15:53,264.651
3-Nov-16,13:16:03,264.58
3-Nov-16,13:16:13,264.368
3-Nov-16,13:16:23,264.273
3-Nov-16,13:16:33,264.391
11/3/16,1:15:52 PM,10.1
11/3/16,1:16:02 PM,10.1
11/3/16,1:16:12 PM,10.1
11/3/16,1:16:22 PM,10.1
11/3/16,1:16:32 PM,10.1

Answer 1

You can use the 'approx' function. Here is an example with your data:
> input <- read.table(text = "11/3/16,1:15:52 PM,10.1
+
+ 11/3/16,1:16:02 PM,10.1
+
+ 11/3/16,1:16:12 PM,10.1
+
+ 11/3/16,1:16:22 PM,10.1
+
+ 11/3/16,1:16:32 PM,10.1", as.is = TRUE, sep = ',')
>
> # convert the date
> input$time <- as.POSIXct(input$V2, format = "%I:%M:%S %p")
> library(lubridate)
>
> input$newtime <- input$time
>
> first <- read.table(text = "3-Nov-16,13:15:53,264.651
+
+ 3-Nov-16,13:16:03,264.58
+
+ 3-Nov-16,13:16:13,264.368
+
+ 3-Nov-16,13:16:23,264.273
+
+ 3-Nov-16,13:16:33,264.391", as.is = TRUE, sep = ',')
> first$time <- as.POSIXct(first$V2, format = "%H:%M:%S")
>
> # use "approx" to interprete values
> # find values for times in "input" since "first" has different values
> input$result <- approx(first$time,
+ first$V3,
+ xout = input$time,
+ rule = 2
+ )$y
>
>
> input
V1 V2 V3 time newtime result
1 11/3/16 1:15:52 PM 10.1 2017-01-11 13:15:52 2017-01-11 13:15:52 264.6510
2 11/3/16 1:16:02 PM 10.1 2017-01-11 13:16:02 2017-01-11 13:16:02 264.5871
3 11/3/16 1:16:12 PM 10.1 2017-01-11 13:16:12 2017-01-11 13:16:12 264.3892
4 11/3/16 1:16:22 PM 10.1 2017-01-11 13:16:22 2017-01-11 13:16:22 264.2825
5 11/3/16 1:16:32 PM 10.1 2017-01-11 13:16:32 2017-01-11 13:16:32 264.3792
> first
V1 V2 V3 time
1 3-Nov-16 13:15:53 264.651 2017-01-11 13:15:53
2 3-Nov-16 13:16:03 264.580 2017-01-11 13:16:03
3 3-Nov-16 13:16:13 264.368 2017-01-11 13:16:13
4 3-Nov-16 13:16:23 264.273 2017-01-11 13:16:23
5 3-Nov-16 13:16:33 264.391 2017-01-11 13:16:33
>

Answer 2

I apologize that I am bit lazy trying evaluating input as you asked. I am still learning R. I wonder if this piece of code solves your fundamental issue.
The algorithm is simple
I change all time to Unix time which is the number of seconds since epoch.
I use unixtime as x and the second number as y
I create new data points based on the second set of datetime as unixtime.
toUnixTime <- function( dateobj ){
return (as.numeric(as.POSIXct(dateobj, origin="1970-01-01")))
}
toDateTime <- function( unixtime){
return (as.POSIXct(unixtime, origin="1970-01-01"))
}
toUnix <- function( datetime ){
return ( as.numeric(strptime( datetime, "%d-%b-%y,%H:%M:%S") ) )
}
toUnix2 <- function( datetime ){
return ( as.numeric(strptime( datetime, "%m/%d/%y,%I:%M:%S %p") ) )
}
main <- function(){
x <- c(toUnix("3-Nov-16,13:15:53" ),toUnix("3-Nov-16,13:16:03:53"))
y <- c(264.651,264.58)
f <- approxfun(x,y)
f(toUnix2("11/3/16,1:16:02 PM"))
}
main()
This outputs
264.5871 for 11/3/16,1:16:02 PM