Search code examples
rtime-seriesstat

R ts( ) function to specify that data has been collected over 10mins interval in a single day

I have word frequencies observed in a data corpus in a single day, across 10minutes time interval. The data looks something like this:

word 00:00:00 00:11:00 00:22:00 00:33:00 . . . .  23:17:00 23:28:00 23:39:00
Robert  10       20        22       32   . . . .   10        11        10
Harvey  12       35        12       12   . . . .   15        26        9

I want to generate a time series object for this data using R. I know that the ts() function can be used to generate time series object where I can specify the start year and the time-interval being months or quarters. Example:

ts(data, frequency=12, start=c(1946,1))

But, I don't know how can I use ts() to specify that data has been collected over 10minutes time intervals in a single day. Any clue?

Also, as you can see my data has time information in the header. Can that be used to generate time series object in R?

Solution

  • Briefly, your main options are

    1. Use ts(), but read the help page and figure out that you can use frequency= or deltat=. As your data frequency is in fact regular, you could could try something like deltat=24*60/11. [ And looking at your data, it seems to be 11 minutes rather than 10 as your question title suggests. ] ts() is useful if you want to fit ARIMA-class models.

    2. Use the zoo package and its zooreg class and see the documentation in the package vignettes. I find zoo to be more easy to work with.