topic Re: Transpose Data in SAS Procedures

Transpose Data

gobejo — Mon, 02 Mar 2015 22:41:05 GMT

Hi,

I have data that looks like this:

ID	party	filing	dt	count
2115389	bank	original	2012-09-13	1
2115389	bank	continuation	2011-10-06	2
2115389	bank	correction	2008-04-09	3

I want the data to look like this:

ID	party	filing	dt	filing	dt	filing	dt
2115389	bank	original	2012-09-13	continuation	2011-10-06	correction	2008-04-09

Re: Transpose Data

Reeza — Mon, 02 Mar 2015 23:11:12 GMT

Note that variable names in SAS need to be unique.

SAS Learning Module: Reshaping data long to wide using the data step

Re: Transpose Data

Jagadishkatam — Tue, 03 Mar 2015 01:22:24 GMT

Hi,

Please try the below code.

As Reeza rightly suggested the variable names must be unique. However an alternative way is to display the variable names with labels, we could assign the same name as variable label. so the variable names will be different but the labels will be same. Hope it meets your requirement.

data have;

input ID party$ filing :$20. dt:yymmdd10. count;

format dt yymmdd10.;

cards;

2115389 bank continuation 2011-10-06 2

2115389 bank correction 2008-04-09 3

2115389 bank original 2012-09-13 1

;

proc sort data=have;

by id party;

run;

data arays;

retain id party filing1 dt1 filing2 dt2 filing3 dt3;

set have;

array vs(3) $20 filing1 - filing3;

array sv(3) dt1 - dt3;

retain filing1 - filing3 dt1 - dt3;

by id party;

if first.party then do;

do i = 1 to 3;

vs(i)='';

sv(i)=.;

end;

vs(count)=filing;

sv(count)=dt;

format dt1 - dt3 yymmdd10.;

label filing1 ='filing'

filing2 ='filing'

filing3 ='filing'

dt1 ='dt'

dt2 ='dt'

dt3 ='dt';

if last.party;

keep id party filing1-filing3 dt1-dt3;

run;

proc print data=arays label;

var _all_;

run;

Thanks,

Jag

Re: Transpose Data

TomKari — Tue, 03 Mar 2015 03:42:19 GMT

Another approach:

data Inter1;
length Column $8 Value $32;
keep ID party Column Value;
set Have;
Column = "dt"||strip(put(Count,best5.));
Value = put(dt, date.);
output;
Column = "filing"||strip(put(Count,best5.));
Value = filing;
output;
run;

proc transpose data=Inter1 out=Want(drop=_:);
by ID party;
id Column;
var Value;
run;

Re: Transpose Data

art297 — Tue, 03 Mar 2015 05:43:33 GMT

With the macro you can download from: http://www.sascommunity.org/wiki/A_Better_Way_to_Flip_(Transpose)_a_SAS_Dataset

You can transpose your data with the following call:

%transpose(data=have, out=want, by=id party, id=count, var=filing dt)

and, if you have a large file, it will run quite a bit faster than using proc transpose.

Re: Transpose Data

gobejo — Tue, 03 Mar 2015 15:02:06 GMT

Hi Jag,

The requirement changed, and modified the following code you gave me.

Data work.ucc_transpose;

LENGTH name_1 date_1 type_1 name_2 date_2 type_2 name_3 date_3 type_3;

SET founders.ucc_count;

array a(3) $50 name_1 - name_3;

array b(3) $20 type_1 - type_3;

array c(3) date_1 - date_3;

retain name_1 - name_3 type_1 - type_3 date_1 - date_3;

by duns_number;

if first.duns_number then do;

do i = 1 to 3;

a(i)='';

b(i)='';

c(i)=.;

end;

a(count)=name;

b(count)=type;

c(count)=date;

format dt1 - dt3 yymmdd10.;

label type_1 ='filing type 1'

type_2 ='filing type 2'

type_3 ='filing type 3'

date_1 ='UCC Date 1'

date_2 ='UCC Date 2'

date_3 ='UCC Date 3';

if last.duns_number;

keep id secured_party_name name_1-name_3 type_1-type_3 date_1-date_3;

run;

and this is what the error log showed:

Data work.ucc_transpose;
26         LENGTH name_1 dt_1 type_1 name_2 dt_2 type_2 name_3 dt_3 type_3;
                                                                          _
                                                                          22
1          1
           _
           352
ERROR 22-322: Expecting a numeric constant.

ERROR 352-185: The length of numeric variables is 3-8.

27         SET founders.ucc_count;
28         array a(3) $50 name_1 - name_3;
29         array b(3) $20 type_1 - type_3;
30         array c(3) dt_1 - dt_3;
31         retain name_1 - name_3 type_1 - type_3 dt_1 - dt_3;
32         by duns_number secured_party_name;
33         if first.duns_number then do;
34         do i = 1 to 3;
35         a(i)='';
36         b(i)='';
37         c(i)=.;
38         end;
39         end;
40         a(count)=name;
41         b(count)=type;
42         c(count)=date;
43         format dt1 - dt3 yymmdd10.;
44         label type_1 ='filing type 1'
45               type_2 ='filing type 2'
2                                                          The SAS System                               08:53 Tuesday, March 3, 2015

46               type_3 ='filing type 3'
47               dt_1 ='UCC Date 1'
48               dt_2 ='UCC Date 2'
49               dt_3 ='UCC Date 3';
50               if last.duns_number;
51               keep id secured_party_name name_1-name_3 type_1-type_3 date_1-date_3;
52         run;

Re: Transpose Data

gobejo — Tue, 03 Mar 2015 15:03:38 GMT

Hi Arthur,

I do not have access to download or install anything.

Thanks,

Tom

Re: Transpose Data

Jagadishkatam — Tue, 03 Mar 2015 15:05:28 GMT

Please replace LENGTH with RETAIN to avoid the error.

Re: Transpose Data

gobejo — Tue, 03 Mar 2015 15:17:26 GMT

It says the variablesis uninitialized:

proc sort data=founders.ucc_count;
26 by duns_number secured_party_name;
27 run;

NOTE: There were 291 observations read from the data set FOUNDERS.UCC_COUNT.
NOTE: The data set FOUNDERS.UCC_COUNT has 291 observations and 5 variables.
NOTE: PROCEDURE SORT used (Total process time):
      real time           0.00 seconds
      user cpu time       0.01 seconds
      system cpu time     0.00 seconds
      memory              503.37k
      OS Memory           16536.00k
      Timestamp           03/03/2015 10:12:49 AM
      Step Count                        34 Switch Count 31
      Page Faults                       0
      Page Reclaims                     60
      Page Swaps                        0
      Voluntary Context Switches        106
      Involuntary Context Switches      46
      Block Input Operations            0
      Block Output Operations           256

28
29
30
31         Data work.ucc_transpose;
32         Retain name_1 dt_1 type_1 name_2 dt_2 type_2 name_3 dt_3 type_3;
33         SET founders.ucc_count;
34         array a(3) $50 name_1 - name_3;
2                                                          The SAS System                               08:53 Tuesday, March 3, 2015

35         array b(3) $20 type_1 - type_3;
36         array c(3) dt_1 - dt_3;
37         retain name_1 - name_3 type_1 - type_3 dt_1 - dt_3;
38         by duns_number secured_party_name;
39         if first.duns_number then do;
40         do i = 1 to 3;
41         a(i)='';
42         b(i)='';
43         c(i)=.;
44         end;
45         end;
46         a(count)=name;
47         b(count)=type;
48         c(count)=date;
49         format dt1 - dt3 yymmdd10.;
50         label type_1 ='filing type 1'
51               type_2 ='filing type 2'
52               type_3 ='filing type 3'
53               dt_1 ='UCC Date 1'
54               dt_2 ='UCC Date 2'
55               dt_3 ='UCC Date 3';
56               if last.duns_number;
57               keep id secured_party_name name_1-name_3 type_1-type_3 date_1-date_3;
58         run;

NOTE: Variable name is uninitialized.
NOTE: Variable type is uninitialized.
NOTE: Variable date is uninitialized.
NOTE: Variable dt1 is uninitialized.
NOTE: Variable dt2 is uninitialized.
NOTE: Variable dt3 is uninitialized.

Thanks

Re: Transpose Data

Jagadishkatam — Tue, 03 Mar 2015 15:25:24 GMT

I think you do not have the name, type and date variables in the dataset you are using in the set statement. could you please provide the proc contents data to me .

please run

proc contents data=founders.ucc_count;

run;

it will help us to see which variables are there in your dataset.