Date: Tue, 3 Feb 2009 15:05:06 -0500
Reply-To: Arthur Tabachneck <art297@NETSCAPE.NET>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Arthur Tabachneck <art297@NETSCAPE.NET>
Subject: Re: array label question
Hari,
In that case, couldn't you just use a rename statement? R=e.g.,:
data need
(rename=(
all1=_2007_1_1_1
all2=_2007_1_1_2
all3=_2007_1_2_1
all4=_2007_1_2_2
all5=_2007_1_3_1
all6=_2007_1_3_2
all7=_2007_2_1_1
all8=_2007_2_1_2
all9=_2007_2_2_1
all10=_2007_2_2_2
all11=_2007_2_3_1
all12=_2007_2_3_2
all13=_2007_3_1_1
all14=_2007_3_1_2
all15=_2007_3_2_1
all16=_2007_3_2_2
all17=_2007_3_3_1
all18=_2007_3_3_2
all19=_2007_4_1_1
all20=_2007_4_1_2
all21=_2007_4_2_1
all22=_2007_4_2_2
all23=_2007_4_3_1
all24=_2007_4_3_2
all25=_2008_1_1_1
all26=_2008_1_1_2
all27=_2008_1_2_1
all28=_2008_1_2_2
all29=_2008_1_3_1
all30=_2008_1_3_2
all31=_2008_2_1_1
all32=_2008_2_1_2
all33=_2008_2_2_1
all34=_2008_2_2_2
all35=_2008_2_3_1
all36=_2008_2_3_2
all37=_2008_3_1_1
all38=_2008_3_1_2
all39=_2008_3_2_1
all40=_2008_3_2_2
all41=_2008_3_3_1
all42=_2008_3_3_2
all43=_2008_4_1_1
all44=_2008_4_1_2
all45=_2008_4_2_1
all46=_2008_4_2_2
all47=_2008_4_3_1
all48=_2008_4_3_2 ))
;
array all {2007 : 2008,1 : 4,1 : 3,1 : 2} ;
do i = 1 by 1 until (last.indiv_id) ;
set have ;
by indiv_id ;
all{yr,qtr,channel_cd,incentive_type} = contacts ;
end ;
run ;
Art
---------
On Tue, 3 Feb 2009 14:45:36 -0500, Hari Nath <hari_s_nath@YAHOO.COM> wrote:
>Hi Art,
>
>Thanks for your reply. Yes i do have the identical code from your proc
>tranpose, but the problem being it takes 30 minutes using an array, but an
>hour and a half using a transpose for ~200 million record with
><10variables.
>
>On Tue, 3 Feb 2009 14:37:24 -0500, Arthur Tabachneck <art297@NETSCAPE.NET>
>wrote:
>
>>Hari,
>>
>>Are you just trying to accomplish something like the following?:
>>
>>data have ;
>> input Obs INDIV_ID yr qtr channel_Cd
>> Incentive_Type CONTACTS ;
>> labl=catx('_',yr,qtr,channel_Cd,Incentive_Type);
>> cards ;
>> 1 100000100004 2007 1 1 2 1
>> 2 100000100004 2007 1 3 2 7
>> 3 100000100004 2008 3 1 2 1
>> 4 100000100005 2007 1 1 2 1
>> 5 100000100005 2007 1 3 2 7
>> 6 100000100005 2008 2 3
>>2
>> 1
>> 7 100000100005 2008 3 3 1 1
>> 8 100000100005 2008 3 3 2 2
>> 9 100000100007 2007 4 1 2 2
>> 10 100000100007 2008 3 1 2 4
>>;
>>
>>proc transpose data=have out=want;
>> by INDIV_ID;
>> id labl;
>> idlabel labl;
>> var contacts;
>>run;
>>
>>Art
>>-------
>>On Tue, 3 Feb 2009 13:45:37 -0500, Hari Nath <hari_s_nath@YAHOO.COM>
>wrote:
>>
>>>Hi all,
>>>
>>>
>>>Following code gives the desired output except the labels i want to. for
>>>example after transposing the data, for the first record with contacts =
>1
>>>its a combination of yr = 2007, qtr = 1, channel = 1 and incentive = 2.
>So
>>>i would label the variable for the corresponding first record as
>>>2007_1_1_2 and so on. I could get this using proc transpose, but its
>>>taking a long time compared to array as the data is few hundred
millions.
>>>Anybody have an idea how can i achieve this using the following array
>>>method.
>>>
>>>thanks in advance.....
>>>
>>>data have ;
>>> input Obs INDIV_ID yr qtr channel_Cd Incentive_Type CONTACTS ;
>>>cards ;
>>> 1 100000100004 2007 1 1 2 1
>>> 2 100000100004 2007 1 3 2 7
>>> 3 100000100004 2008 3 1 2 1
>>> 4 100000100005 2007 1 1 2 1
>>> 5 100000100005 2007 1 3 2 7
>>> 6 100000100005 2008 2 3 2 1
>>> 7 100000100005 2008 3 3 1 1
>>> 8 100000100005 2008 3 3 2 2
>>> 9 100000100007 2007 4 1 2 2
>>> 10 100000100007 2008 3 1 2 4
>>>;
>>>run ;
>>>
>>>data need ;
>>> array all {2007 : 2008,1 : 4,1 : 3,1 : 2} ;
>>>
>>> do i = 1 by 1 until (last.indiv_id) ;
>>> set have ;
>>> by indiv_id ;
>>>
>>> all{yr,qtr,channel_cd,incentive_type} = contacts ;
>>> end ;
>>>run ;
| 
