R help - Nov 2010 - Counting things in a time series.

Hi,

I have a process (not in R) that records events with a time stamp.  So,
I have a huge series of maybe 100,000 time stamps.

I'd like to break it up into hourly (Or daily) intervals and then count
how many events occurred in each interval.  That way I can graph it.

Ideally, converting the this into a time series in R would let me do
some interesting analysis.

The data is just a list of epoch timestamps.  Importing into R is
trivial but I'm stuck from there.

1) How can I "bin" the counts by an hour?  One thought would be to
divided each timestamp by 3600, them multiply back by 1000.  This would
effectively convert them to hours with multiple entries per hour. ) 
But, I then don't know how to count them

2) Once I figure out the counts, I'll then have a data structure with a
column of epoch seconds and a second column of counts.  How can I then
convert that into a ts object?

Thanks!

-N

You never did show us what your data looks like.  You could convert
them to POSIXct, then use 'cut' to split them into the bins and then
'table' to count them.

Try something like this:
> timeStamp <- as.POSIXct('2010-11-21 00:00') + runif(50, 0,
86400)  # day's worth of time
> # bin into 1 hour buckets starting at midnight
> bins <- cut(timeStamp, breaks = seq(as.POSIXct('2010-11-21
00:00'),
+                             by = '1 hour', length =
25))
>
> table(bins)
bins
2010-11-21 00:00:00 2010-11-21 01:00:00 2010-11-21 02:00:00 2010-11-21
03:00:00 2010-11-21 04:00:00
                  1                   5                   4
       5                   5
2010-11-21 05:00:00 2010-11-21 06:00:00 2010-11-21 07:00:00 2010-11-21
08:00:00 2010-11-21 09:00:00
                  2                   2                   4
       1                   3
2010-11-21 10:00:00 2010-11-21 11:00:00 2010-11-21 12:00:00 2010-11-21
13:00:00 2010-11-21 14:00:00
                  0                   3                   3
       0                   1
2010-11-21 15:00:00 2010-11-21 16:00:00 2010-11-21 17:00:00 2010-11-21
18:00:00 2010-11-21 19:00:00
                  0                   3                   0
       3                   1
2010-11-21 20:00:00 2010-11-21 21:00:00 2010-11-21 22:00:00 2010-11-21 23:00:00
                  0                   1                   1                  
2
>

On Sun, Nov 21, 2010 at 12:29 AM, Noah Silverman
<noah at smartmediacorp.com> wrote:
> Hi,
>
> I have a process (not in R) that records events with a time stamp. ?So,
> I have a huge series of maybe 100,000 time stamps.
>
> I'd like to break it up into hourly (Or daily) intervals and then count
> how many events occurred in each interval. ?That way I can graph it.
>
> Ideally, converting the this into a time series in R would let me do
> some interesting analysis.
>
> The data is just a list of epoch timestamps. ?Importing into R is
> trivial but I'm stuck from there.
>
> 1) How can I "bin" the counts by an hour? ?One thought would be
to
> divided each timestamp by 3600, them multiply back by 1000. ?This would
> effectively convert them to hours with multiple entries per hour. )
> But, I then don't know how to count them
>
> 2) Once I figure out the counts, I'll then have a data structure with a
> column of epoch seconds and a second column of counts. ?How can I then
> convert that into a ts object?
>
> Thanks!
>
> -N
>
> ______________________________________________
> 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.
>


-- 
Jim Holtman
Cincinnati, OH
+1 513 646 9390

What is the problem that you are trying to solve?

Hi, these are pretty basic questions. You might want to pick up an
introductory manual.

Lets assume you have a time stamp that already indicates the hours. Assume
you have 300 observations, each of which falls in one of 24 hours of
observation. You easily get the number of obs in each hour with the table
command.

x=sample(1:24,300,replace=T)
table(x)

Let's now assume you don't have the observations binned yet. Then you
can
'cut' the time stamp vector at intervals of your liking, and tabulate
the
resulting occurrences of the factors. Assume as an example that you wanted
irregular intervals. You want to know how many occurrences there are between
midnight and 0900, 0900 and noon, noon and 1500, 1500 and 1800, and 1800 and
midnight.

nx=cut(x,c(0,9,12,15,18,24))
table(nx)

for the second question: ?as.ts()

HTH,
Daniel

-- 
View this message in context:
http://r.789695.n4.nabble.com/Counting-things-in-a-time-series-tp3052148p3052306.html
Sent from the R help mailing list archive at Nabble.com.