Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: Testing for a contiguous subset of observations?


From   BW Wheeler <ben.wheeler@bristol.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   RE: st: Testing for a contiguous subset of observations?
Date   Fri, 30 Nov 2007 12:18:23 -0000

Thanks Maarten & Nick -

Nick you are right, Maarten's method produces too many missing values; I should have specified that there would be multiple spells of zeros.

I guess 'spells' was the magic search word that I was missing! tsspell and the references you mention look like they'll do the job.

Thanks a lot
Ben

tsspell and the associated material looks like it will do the job -

--On 30 November 2007 12:01 +0000 Nick Cox <n.j.cox@durham.ac.uk> wrote:


This lumps together all the zeros for each -id- regardless of when they
occurred.
That is valid if there is at most one spell of zeros for each -id-.
Otherwise
I don't think it matches the problem.

Nick
n.j.cox@durham.ac.uk

Maarten buis

bys id (observation) : gen zero = Activity == 0
bys id zero (observation) : gen nzero = _N if zero == 0
replace nzero = 0 if nzero == .
gen activity2 = activity if nzero < 60 & nzero

--- BW Wheeler <ben.wheeler@bristol.ac.uk> wrote:

I'm having trouble figuring out a solution to this data management
problem, and have exhausted all books and findit (partly as it's hard
to know what to search for)!

I have time series data captured from accelerometers, with an
observation of 'activity counts' logged every 10 seconds for several
days for each individual. If the number of counts is zero continuously

for any period of 10 minutes or more (i.e. 60+ observations), we need
to assume that the accelerometer was not being worn, and to treat
these zeros as missing.

So, my data look something like this:

Observation		Date		Time		Activity
1			13/05/2006	15:20:00	230
2			13/05/2006	15:20:10	215
3			13/05/2006	15:20:20	0
4			13/05/2006	15:20:30	0
5			13/05/2006	15:20:40	0
...

Does anyone have any ideas of how I can replace the 'Activity' value
with missing, if that value is a zero in the midst of a continuous
series of 60+ zeros?
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


----------------------
Ben Wheeler
Research Fellow
Department of Social Medicine, University of Bristol
& South West Public Health Observatory
ben.wheeler@bristol.ac.uk

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index