[R] Inserting rows of interpolated data

[Joshua Ulrich]

Hi Jon,

zoo is great for tasks like this, not just for na.approx. :)

I would approach the problem like this:

library(zoo)
# put lightdata into a zoo object
z <- with(lightdata, zoo(light,
  as.POSIXct(paste(date, time), format="%m/%d/%y %H:%M:%S")))
# merge the above zoo object with an "empty" zoo
# object that has all the index values you want
Z <- merge(z, zoo(,seq(start(z),end(z),by="1 min")))
# interpolate between the 5-min observatoins
Z <- na.approx(Z)

HTH,
--
Joshua Ulrich  |  about.me/joshuaulrich
FOSS Trading  |  www.fosstrading.com

R/Finance 2013: Applied Finance with R  | www.RinFinance.com


On Mon, Feb 11, 2013 at 5:19 PM, Benstead, Jonathan <jbenstead at as.ua.edu> wrote:
> Dear help list - I have light data with 5-min time-stamps. I would like to insert four 1-min time-stamps between each row and interpolate the light data on each new row. To do this I have come up with the following code:
>
> lightdata <- read.table("Test_light_data.csv", header = TRUE, sep = ",")  # read data file into object "lightdata"
> library(chron)
> mins <- data.frame(times(1:1439/1440)) # generate a dataframe of 24 hours of 1-min timestamps
> Nth.delete <- function(dataframe, n)dataframe[-(seq(n, to=nrow(dataframe), by=n)),] # function for deleting nth row
> empty <- data.frame("1/9/13", Nth.delete(mins, 5), "NA") # delete all 5-min timestamps in a new dataframe
> colnames(empty) <- c("date", "time", "light") # add correct column name to empty timestamp dataframe
> newdata <- rbind(lightdata, empty)
>
> I get the following error message:
>
> Warning message:
> In `[<-.factor`(`*tmp*`, ri, value = c(0.000694444444444444, 0.00138888888888889,  :
>   invalid factor level, NAs generated
>
> Digging into this a little, I can see that the two time columns are doing what I need and APPEAR to be similar in format:
>
>> head(lightdata)
>     date    time       light
> 1 1/9/13 0:00:00 -0.00040925
> 2 1/9/13 0:05:00 -0.00023386
> 3 1/9/13 0:10:00 -0.00032155
> 4 1/9/13 0:15:00 -0.00017539
> 5 1/9/13 0:20:00 -0.00029232
> 6 1/9/13 0:25:00 -0.00038002
>
>> head(empty)
>     date     time light
> 1 1/9/13 00:01:00    NA
> 2 1/9/13 00:02:00    NA
> 3 1/9/13 00:03:00    NA
> 4 1/9/13 00:04:00    NA
> 5 1/9/13 00:06:00    NA
> 6 1/9/13 00:07:00    NA
>
> but they clearly are not as far as R is concerned, as shown by str:
>
>> str(lightdata)
> 'data.frame':   288 obs. of  3 variables:
>  $ date : Factor w/ 1 level "1/9/13": 1 1 1 1 1 1 1 1 1 1 ...
>  $ time : Factor w/ 288 levels "0:00:00","0:05:00",..: 1 2 3 4 5 6 7 8 9 10 ...
>  $ light: num  -0.000409 -0.000234 -0.000322 -0.000175 -0.000292 ...
>
>> str(empty)
> 'data.frame':   1152 obs. of  3 variables:
>  $ date : Factor w/ 1 level "1/9/13": 1 1 1 1 1 1 1 1 1 1 ...
>  $ time :Class 'times'  atomic [1:1152] 0.000694 0.001389 0.002083 0.002778 0.004167 ...
>   .. ..- attr(*, "format")= chr "h:m:s"
>  $ light: Factor w/ 1 level "NA": 1 1 1 1 1 1 1 1 1 1 ...
>
> In the first (original) dataframe, light is a factor, while in the dataframe of generated timestamps, the timestamps are actually still in fractions of a day.
>
> Presumably this is why rbind is not working? Can anyone help? By the way, I know I can use na.approx in zoo to do the eventual interpolation of the light data. It's getting there that has me stumped for now.
>
> Many thanks, Jon (new R user).
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.