Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: add observations
From
Nick Cox <[email protected]>
To
[email protected]
Subject
Re: st: add observations
Date
Mon, 6 Aug 2012 17:35:28 +0100
The rules for your data remain cryptic to me. I think you are saying
that for at least some values of -hcod- only three distinct values of
-pd- are observed and the other one is the appropriate replacement.
You could try something like this
gen obs = ""
bysort hcod (pd) : replace obs = obs + string(pd) if _n == 1 | pd != pd[_n-1]
gen rep = 4 if obs == "123"
replace rep = 3 if obs = "124"
replace rep = 2 if obs = "134"
replace rep = 1 if obs = "234"
local N = _N + 1
expand 2
replace pd = rep in `N'/L
On Mon, Aug 6, 2012 at 4:21 PM, Ronnie Babigumira <[email protected]> wrote:
> Thanks Nick,
> This works perfectly for the one unit, however, I have many units and, the missing pd varies from unit to unit. Suggestions on how your solution can be modified to handle this?
>
> My early thinking (and block) was thinking I needed to used -fillin- (and may be levels) however, after sending the email to the list it occurred to me that splitting and then appending the data may do the job. Here is what I have tried.
>
> clear
> input ///
> str8 hcod pd pdt
> 2081115 1 12
> 2081115 3 21
> 2081115 3 11
> 2081115 4 13
> 3031028 1 45
> 3031028 2 21
> 3031028 2 11
> 3031028 3 42
> 3031028 3 13
> 3031028 3 12
> 3061037 2 12
> 3061037 3 45
> 3061096 1 42
> 3061096 1 21
> 3061096 4 45
> 3061096 4 12
> 3061196 1 42
> 3061196 1 45
> 3061196 4 12
> 3061306 1 21
> 3061306 1 45
> 3061306 4 42
> 3061306 4 12
> 3062277 1 41
> 3062277 4 12
>
>
> end
>
> tempfile master
> save `master'
>
> fillin hcod pd
> bysort hcod (pdt): replace pd = pd[_N]
> drop if mi(pdt)
> append using `master'
> sort hcode pd
> list, clean noobs
>
> Ronnie
> --
> 010100100110111101101110011011100110100101100101
>
>
>
>
> On Monday, August 6, 2012 at 3:07 PM, Nick Cox wrote:
>
>> local N = _N + 1
>> expand 2
>> replace pd =4 in `N'/L
>>
>>
>> On Mon, Aug 6, 2012 at 1:25 PM, Ronnie Babigumira <[email protected] (mailto:[email protected])> wrote:
>> > Dear listers,
>> >
>> > I have a dataset that looks like this
>> > hcod pd pdt
>> > 3092285 1 31
>> > 3092285 1 43
>> > 3092285 2 11
>> > 3092285 2 21
>> > 3092285 2 41
>> > 3092285 2 45
>> > 3092285 3 42
>> >
>> >
>> >
>> >
>> > I would like to add new observations so that I end up with
>> > hcod pd pdt
>> > 3092285 1 31
>> > 3092285 1 43
>> > 3092285 2 11
>> > 3092285 2 21
>> > 3092285 2 41
>> > 3092285 2 45
>> > 3092285 3 42
>> > 3092285 4 31
>> > 3092285 4 43
>> > 3092285 4 11
>> > 3092285 4 21
>> > 3092285 4 41
>> > 3092285 4 45
>> >
>> > 3092285 4 42
>> >
>> >
>> >
>> > Basically, each unit (hcod) should have 4 periods (pd). What we would like to do is add all products (pdt) present in the non-missing periods to the missing period.
>> >
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/