retaining
in [SAS]
|
From: Ya Huang on 2 Jan 2007 12:20 Happy new year Venkey, Your code may have problem (depends on what the original poster wants). I changed the input data, so that there is a missing after the first non-missing: 2 20060502 2 20060627 3 . 3 . 3 20070101 3 20070102 3 . <- missing after first nonmissing; 3 20070102 run ; The result is like this: 2 20060627 3 20070101 3 20070101 3 20070101 3 20070102 3 20070101 <- I guess it would make more sense if this is 20070102 ? 3 20070102 On Tue, 2 Jan 2007 12:02:36 -0500, Venky Chakravarthy <swovcc(a)HOTMAIL.COM> wrote: >On Tue, 2 Jan 2007 00:38:52 -0800, deepalipowale(a)GMAIL.COM wrote: > >>hi all , >> >>i want to retained FIRST non-missing values. >>e.g. ORIGINAL DATASET >> >>ID DATE >>1 20051018 >>1 20051213 >>2 >>2 20060502 >>2 20060627 >> >>I want it like this: >> >>REQUIRED >> >>ID DATE NEW_DATE >>1 20051018 20051018 >>1 20051213 20051018 >>2 20060502 >>2 20060502 20060502 >>2 20060627 20060502 > >I am assuming that within an ID, once a date becomes non-missing it is >always present after that. Note that your example is expanded slightly to >accommodate more than one missing date within an ID but it is consecutive. >Also note that if you have missing dates between valid dates it will be >filled with the first non-missing date value for that ID. > >The modification to the classical DoW stops processing the first SET after >the first non-missing date observation is read. That is all that is needed >from it. The second SET outputs all observations but only after it fills >missing DATE values with FIRSTDATE. > >data withmissing ; > input ID DATE : yymmdd. ; > format date yymmddn8. ; > cards ; >1 20051018 >1 20051213 >2 . >2 20060502 >2 20060627 >3 . >3 . >3 20070101 >3 20070102 >run ; > >data filled ; > do until (first.id) ; > set withmissing (where = (firstdate^=.) rename = (date=firstdate)); > by id ; > end ; > do until (last.id) ; > set withmissing ; > by id ; > if missing(date) then date = firstdate ; > output ; > end ; > drop firstdate ; >run ; > >Venky Chakravarthy
From: Venky Chakravarthy on 2 Jan 2007 14:06 Hi Ya, A very happy new year to you too. You are indeed correct that the code would have problems with the example that you presented. Those were exactly my thoughts when I first looked at the problem. I tried to address those points in my first paragraph which stated: >>Also note that if you have missing dates between valid dates it will be >>filled with the first non-missing date value for that ID. But also bear in mind that the original poster's request was: >>>i want to retained FIRST non-missing values. So the burden is on the original poster to clarify how he/she wants it to be processed when there is a missing value between valid values. Would that still need to be filled with the FIRST non-missing value or the NEXT non- missing value? Regards, Venky On Tue, 2 Jan 2007 12:20:26 -0500, Ya Huang <ya.huang(a)AMYLIN.COM> wrote: >Happy new year Venkey, > >Your code may have problem (depends on what the original poster wants). >I changed the input data, so that there is a missing after the first >non-missing: > >2 20060502 >2 20060627 >3 . >3 . >3 20070101 >3 20070102 >3 . <- missing after first nonmissing; >3 20070102 >run ; > >The result is like this: > > 2 20060627 > 3 20070101 > 3 20070101 > 3 20070101 > 3 20070102 > 3 20070101 <- I guess it would make more sense if this is 20070102 ? > 3 20070102 > > > >On Tue, 2 Jan 2007 12:02:36 -0500, Venky Chakravarthy <swovcc(a)HOTMAIL.COM> >wrote: > >>On Tue, 2 Jan 2007 00:38:52 -0800, deepalipowale(a)GMAIL.COM wrote: >> >>>hi all , >>> >>>i want to retained FIRST non-missing values. >>>e.g. ORIGINAL DATASET >>> >>>ID DATE >>>1 20051018 >>>1 20051213 >>>2 >>>2 20060502 >>>2 20060627 >>> >>>I want it like this: >>> >>>REQUIRED >>> >>>ID DATE NEW_DATE >>>1 20051018 20051018 >>>1 20051213 20051018 >>>2 20060502 >>>2 20060502 20060502 >>>2 20060627 20060502 >> >>I am assuming that within an ID, once a date becomes non-missing it is >>always present after that. Note that your example is expanded slightly to >>accommodate more than one missing date within an ID but it is consecutive. >>Also note that if you have missing dates between valid dates it will be >>filled with the first non-missing date value for that ID. >> >>The modification to the classical DoW stops processing the first SET after >>the first non-missing date observation is read. That is all that is needed >>from it. The second SET outputs all observations but only after it fills >>missing DATE values with FIRSTDATE. >> >>data withmissing ; >> input ID DATE : yymmdd. ; >> format date yymmddn8. ; >> cards ; >>1 20051018 >>1 20051213 >>2 . >>2 20060502 >>2 20060627 >>3 . >>3 . >>3 20070101 >>3 20070102 >>run ; >> >>data filled ; >> do until (first.id) ; >> set withmissing (where = (firstdate^=.) rename = (date=firstdate)); >> by id ; >> end ; >> do until (last.id) ; >> set withmissing ; >> by id ; >> if missing(date) then date = firstdate ; >> output ; >> end ; >> drop firstdate ; >>run ; >> >>Venky Chakravarthy
From: "Howard Schreier <hs AT dc-sug DOT org>" on 2 Jan 2007 23:46 On Tue, 2 Jan 2007 01:01:24 -0800, Mogens A. Krogh <MKROGH(a)DSR.KVL.DK> wrote: >You could do something like this, if your data is sorted in ID DATE > >data test; >input ID $ DATE ; >cards; >1 20051018 >1 20051213 >2 . >2 20060502 >2 20060627 >;; >run; >data want; >merge test test (rename=(date=_newdate) where=(_newdate ^= .)); >by ID; >if first.id then newdate=_newdate; >retain newdate; >drop _:; >run; Or data want; do until(last.id); set test test(in=secondpass); by ID; if missing(new_date) then new_date = date; if secondpass then output; end; run; > >regards >Mogens A. Krogh >DVM, PhD-student >Faculty of Life Sciences >University of Copenhagen, Denmark > > >deepalipowale(a)gmail.com skrev: >> hi all , >> >> i want to retained FIRST non-missing values. >> e.g. ORIGINAL DATASET >> >> ID DATE >> 1 20051018 >> 1 20051213 >> 2 >> 2 20060502 >> 2 20060627 >> >> I want it like this: >> >> REQUIRED >> >> ID DATE NEW_DATE >> 1 20051018 20051018 >> 1 20051213 20051018 >> 2 20060502 >> 2 20060502 20060502 >> 2 20060627 20060502
First
|
Prev
|
Pages: 1 2 Prev: SAS message Log: Generic critical error. Next: proc surveyselect problem |