[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
RE: st: On the analysis of intraday data
William Gould replied to Levy Lee
> > [...] I don't know if stata has some builtin functions to
> handle time
> > variables, as it does on date variables. I have a variables
> looks like the
> > following:
> > Obs. No. Time
> > 1 2003-03-03 09:30:29
> > 2 2003-03-03 09:30:47
> > 3 2003-03-03 09:31:11
> > [...] I wish there is some buildin function so that we can
> > 2003-03-03 09:30:29 minus 2003-03-03 09:30:47 will equals
> to 28 seconds.
> There is no built-in function, but obtaining the desired
> result should not
> be too difficult. Although this is "intraday" data, let's
> write our code
> so that, were times interday, we would still get the right answer.
> 2003-03-03 09:30:29
> . gen str date = substr(time, 1, 10)
> . assert substr(time,11,1)==" "
> . gen hour = real(substr(time,12,2))
> . assert substr(time,14,1)==":"
> . gen min = real(substr(time,15,2))
> . assert substr(time,17,1)==":"
> . gen sec = real(substr(time,18,2))
> Now that I've got the components -- and I have established that all
> observations have the expected format, we can make the calculation:
> . gen edate = date(date, "ymd")
> . gen double secs = date*24*60*60 + hour*60*60 + min*60 + sec
> Variable -secs- now contains the number of seconds since
> 01jan1960 00:00:00.
> Note that I have stored -secs- as a double. That is important.
It may be of interest to spell out the similarity between
this solution and one using -split-. -substr()- and -real()-
and -date()- are all important functions that almost everyone
needs eventually in their Stata work. -split- is by
contrast a convenience command based on -substr()-, so we
are talking about the same solution, approached from zeroth
principles or a convenience command.
In the case of -time-, with values like "2003-03-03 09:30:29"
we want to -split- on the space and the colons, and we want
to get numeric values where possible, so
. split time, p(" " :) destring
This should yield
time1 day (cannot -destring-)
time2 hours (should be able to -destring-)
time3 minutes (same)
time4 seconds (same)
Look at -split-'s output for signs of problems,
or use -assert- to test.
Then the last steps are very similar:
. gen edate = date(time1, "ymd")
. gen double secs = edate*24*60*60 + time2*60*60 + time3*60 + time4
The last corrects a typo in Bill's last line ("date" should be
* For searches and help try: