topic Re: Creating new variable from data in multiple variables in SAS Programming

Creating new variable from data in multiple variables

kk11 — Sun, 13 Jan 2019 19:29:19 GMT

Hi,

I tried the following code to create a new variable from existing multiple variables using ARRAY statement and DO OVER loop. But I am not able to get the right result.

My data has variables DX1, DX2, DX3... DX30. These variables are character. Data in these variables looks like,

DX1	DX2	DX3	DX4	DX5	DX6	DX7	DX8	DX9	DX10	DX11	DX12	DX13	DX14	DX15	DX16	DX17	DX18	DX19	DX20	DX21	DX22	DX23	DX24	DX25	DX26	DX27	DX28	DX29	DX30
4241	42832	20410	4280	V707	4019	30391	42789	78194	V1582
4241	4260	496	4019	2724	4439	V707	42731	41401	V1254	32723	V462	V5861	V4582	V1582
4241	78551	42822	5853	2875	4168	4280	2851	V707	496	2724	41400	27800	40390	44020
4241	42832	4280	25000	2720	2724	42731	43310	40390	5859	4168	311	V5866	V5867	V5869
4241	42822	4260	4254	4280	V1582	496	4019	2724	25000	60000	V1083	V1254	27800	V1582
3950	41071	3910	39891	2536	41401	4019	V4582	42731	340	59654	4168	4280	2724	43490
99602	42830	4280	2875	25040	4241	V422	41400	V1582	V4582	V707	V5866	40390	5859	58381
4241	51882	V707	4280	42611	4263	41401	4142	5859	40390	79311	2749	7993	33818	4928	25000	4439	53081	60000	2724	412	V5863	V4582	V1582
4241	V707	42843	4168	V462	V8543	4280	41401	27801	496	32723	2449	25000	4019	6202	6202	1103	412	V5866	V5863	V173	V5867	V1582
4241	5849	45341	42832	4280	2875	78630	V707	4263	42731	4019	2724	41400	V4581	V5861	42781	797	V5866	V1582

Some of the values in these variables like - "V1582","3051 ","30510","30511","30512","30513","64900","64901","64902","64903","64904" - represent one condition. I want to make a new variable with coding "1" if the observations in the variables from DX1 to DX30 have the values I mentioned in the prior sentence and coding "0" if not.

I used the following code, but I am not getting the right result.

Data set has 42690044 observations.

DATA PROJECT.SAMPLE;
SET PROJECT.SAMPLE;
ARRAY VARIABLE $ DX1-DX30;
DO OVER VARIABLE;
IF (VARIABLE) in ("V1582","3051","30510","30511","30512","30513","64900","64901","64902","64903","64904") THEN NEWDX=1; ELSE NEWDX=0;
END;
RUN;

The frequency of the new variable is,

(NEWDX = 0) ==> 42687442 (99.99 %)

(NEWDX = 1) ==> 2602 (0.01%)

Actually I added the frequencies of each of these valves - "V1582","3051 ","30510","30511","30512","30513","64900","64901","64902","64903","64904" - in all the variables DX1 to DX30 and I am getting around ,

(NEWDX = 0) ==> 72 %

(NEWDX = 1) ==> 28%.

The observations with new condition should be 28%, but when I use the my code it is giving only 0.01%.

So my code is not doing exactly what I want. But the code worked for few observation like 100 in a small sample data set. I am not sure what is going wrong when I apply to large data set.

And I am not getting any error.

Can any one please help me to accomplish my task.

Thank you,

Kamesh.

Re: Creating new variable from data in multiple variables

novinosrin — Sun, 13 Jan 2019 19:40:05 GMT

Occasional learner and do over implicit array, thats interesting

Try this

 

DATA PROJECT.SAMPLE;
SET PROJECT.SAMPLE;
ARRAY VARIABLE $ DX1-DX30;
NEWDX=0;
DO OVER VARIABLE;
IF (VARIABLE) in ("V1582","3051","30510","30511","30512","30513","64900","64901","64902","64903","64904") 
THEN do;
NEWDX=1;  
leave;
end;
END;
RUN

Re: Creating new variable from data in multiple variables

kk11 — Sun, 13 Jan 2019 19:55:41 GMT

Thank you for responding.

I tried the code you mentioned, but I got error.

1 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;
61
62 DATA PROJECT.SAMPLE;
63 SET PROJECT.SAMPLE;
64 ARRAY VARIABLE $ DX1-DX30;
65 NEWDX=0;
66 DO OVER VARIABLE;
67 IF (VARIABLE) in
67 ! ("V1582","3051","30510","30511","30512","30513","64900","64901","64902","64903","64904")
68 THEN do;
69 NEWDX=1;
70 leave;end;
71 END;
72 RUN
73
74 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;
_______
22
76
ERROR 22-322: Syntax error, expecting one of the following: ;, CANCEL, PGM.

ERROR 76-322: Syntax error, statement will be ignored.

75 ODS HTML CLOSE;
76 &GRAPHTERM; ;*';*";*/;RUN;QUIT;
76 &GRAPHTERM; ;*';*";*/;RUN;QUIT;
_
180
ERROR 180-322: Statement is not valid or it is used out of proper order.

NOTE: The SAS System stopped processing this step because of errors.
WARNING: The data set PROJECT.SAMPLE may be incomplete. When this step was stopped there were 0
observations and 347 variables.
NOTE: DATA statement used (Total process time):
real time 0.76 seconds
cpu time 0.01 seconds

77 QUIT;RUN;
78 ODS HTML5 (ID=WEB) CLOSE;
79
80 ODS RTF (ID=WEB) CLOSE;
81 ODS PDF (ID=WEB) CLOSE;
NOTE: ODS PDF(WEB) printed no output.
(This sometimes results from failing to place a RUN statement before the ODS PDF(WEB) CLOSE
statement.)
82 FILENAME _GSFNAME;
NOTE: Fileref _GSFNAME has been deassigned.

83 DATA _NULL_;
84 RUN;

NOTE: DATA statement used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

85 OPTIONS VALIDMEMNAME=COMPAT;
86 OPTIONS NOTES STIMER SOURCE SYNTAXCHECK;
87

Re: Creating new variable from data in multiple variables

novinosrin — Sun, 13 Jan 2019 20:00:22 GMT

missed a semicolon after RUN

DATA PROJECT.SAMPLE;
SET PROJECT.SAMPLE;
ARRAY VARIABLE $ DX1-DX30;
NEWDX=0;
DO OVER VARIABLE;
IF (VARIABLE) in ("V1582","3051","30510","30511","30512","30513","64900","64901","64902","64903","64904") 
THEN do;
NEWDX=1;  
leave;end;
END;
RUN;

Re: Creating new variable from data in multiple variables

kk11 — Sun, 13 Jan 2019 20:03:55 GMT

Sorry, I didn't realize. Was excited with your reply and copied and pasted the code. Thank you.

Re: Creating new variable from data in multiple variables

novinosrin — Sun, 13 Jan 2019 20:06:44 GMT

No worries and it was my mistake too. Sorry about that. Let us know if that works

Re: Creating new variable from data in multiple variables

kk11 — Sun, 13 Jan 2019 20:29:46 GMT

Thank you very much.

You solved my problem. It worked.

I am a new learner. Trying to understand more about IF THEN and ELSE statement.

I have couple questions.

I tried my code without using " ELSE NEWDX=0;" after "IF compress(DIAGNOSIS) in ("V1582","3051","30510","30511","30512","30513","64900","64901","64902","64903","64904") THEN Smoking=1;".

Like below,

DATA PROJECT.SAMPLE;
SET PROJECT.SAMPLE;
ARRAY VARIABLE $ DX1-DX30;
DO OVER VARIABLE;
IF compress(VARIABLE) in ("V1582","3051","30510","30511","30512","30513","64900","64901","64902","64903","64904") THEN NEWDX=1;
END;
RUN;

It worked, but it coded NEWDX as 1 if the "IF THEN" statement is satisfied and left remaining blank (missing data).

Frequency table

NEWDX Frequency Percent

. 7472 .

1 3061 100

After I used your code,

I got similar result, but it coded "0" fro remaining observations.

Frequency table

NEWDX Frequency Percent

0 7472 70.94

1 3061 29.06

When I used " ELSE NEWDX=0;"

after the

"IF compress(DIAGNOSIS) in ("V1582","3051","30510","30511","30512","30513","64900","64901","64902","64903","64904") THEN Smoking=1;"

I am getting

NEWDX Frequency Percent

0 10527 99.94

1 6 0.06

So, using ELSE making my result wrong.

Now I learned that I am using ELSE in wrong context.

Can you explain me why ELSE is not working?

I understood your code little bit, but can you please explain me your code and suggest any good article that will give me better understanding about DO loops and will help me to make good code like you did.

I really appreciate your help.

Thank you very much.

Re: Creating new variable from data in multiple variables

novinosrin — Sun, 13 Jan 2019 20:47:55 GMT

Ok,

Let's take your 1st conditional statement

IF (VARIABLE) in ("V1582","3051","30510","30511","30512","30513","64900","64901","64902","64903","64904") THEN NEWDX=1; ELSE NEWDX=0;

1. Newdx var is assigned a blank(missing) value at the top of each iteration of the datastep besides at compile time.

2. When you loop hrough the elements of the array, you will need to exit the loop soon as NEWDX=1 otherwise the loop continues and execute and if the next element happens to be not true(newdx=1) then else condition takes effect and overwrites the NEWDX=0, which is not desired. That's what I modified with leave statement.

3. So for all the other values where NEWDX=1 is not true, the blank(missing) value initialised at compile time remains in effect and we have to find a way to make zero.

4. So assigning NEWDX=0 at te top makes sure this is done and, when and if the conditional statement in the loop is true, NEWDX=0 is replcaed by 1.

5. You can also use boolean expressions like


do over variable;
Newdx=(VARIABLE) in ("V1582","3051","30510","30511","30512","30513","64900","64901","64902","64903","64904");
if newdx then leave;
end;

This is just a fancy

HTH

Re: Creating new variable from data in multiple variables

kk11 — Sun, 13 Jan 2019 20:46:28 GMT

I have another question,

If I want to create a new varaible in the same way this time from two ARRAY's. The second ARRAY has same number of varaibels as first ARRAY and named as CHRON1, CHRON2, CHRON3, ............CHRON30.

CHRON1 Varaible is binary and coded as "1" and "0" and it corresponds to DX1 and represents whether vakve in DX1 is chronic or not.

CHRON1 is numeric.

If I want to satisfy both the values in DX1 and CHRON1 to cerate new varaible "NEWDX", can I use the "AND" in the "IF THEN statement" like below?

Should I have to mention the ARRAY name "CHRONIC" in DO OVER statement along with ARRAY name "VARIABLE" - "DO OVER VARIABLE AND CHRONIC"

can I just leave the "DO OVER VARIABLE" like below and it automatically considers the ARRAY "CHRONIC" ?

DATA PROJECT.SAMPLE;
SET PROJECT.SAMPLE;
ARRAY VARIABLE $ DX1-DX30;
ARRAY CHRONIC CHRON1-CHRON30;
NEWDX=0;
DO OVER VARIABLE;
IF (VARIABLE) in ("438","4380","43810","43811","43812","43813","43814","43819","43820",) AND (CHRONIC) in ('1')
THEN do;
NEWDX=1;
leave;end;
END;
RUN;

Please let me know.

Thank you.

Re: Creating new variable from data in multiple variables

kk11 — Sun, 13 Jan 2019 20:54:31 GMT

Your explanation is really helpful. Now I understood.

Thank you very much for teaching me.

Please let me know if there are any corrections I have to make in the code I posed using 2 ARRAY's as my condition need to satisfy the values in two different series of variables.

Re: Creating new variable from data in multiple variables

kk11 — Sun, 13 Jan 2019 20:55:33 GMT

I have another question,

If I want to create a new varaible in the same way this time from two ARRAY's. The second ARRAY has same number of varaibels as first ARRAY and named as CHRON1, CHRON2, CHRON3, ............CHRON30.

CHRON1 Varaible is binary and coded as "1" and "0" and it corresponds to DX1 and represents whether vakve in DX1 is chronic or not.

CHRON1 is numeric.

If I want to satisfy both the values in DX1 and CHRON1 to cerate new varaible "NEWDX", can I use the "AND" in the "IF THEN statement" like below?

Should I have to mention the ARRAY name "CHRONIC" in DO OVER statement along with ARRAY name "VARIABLE" - "DO OVER VARIABLE AND CHRONIC"

can I just leave the "DO OVER VARIABLE" like below and it automatically considers the ARRAY "CHRONIC" ?

Please let me know.

Thank you.

Re: Creating new variable from data in multiple variables

Allaluiah — Sun, 13 Jan 2019 21:48:40 GMT

@kk11 Hi, Would be more appropriate and courteous to mark novinosrin's solution as accepted than marking your own. Thanks!

Re: Creating new variable from data in multiple variables

kk11 — Sun, 13 Jan 2019 21:57:25 GMT

Sorry, I am new and didn't know. I will mark novinosrin's solution as accepted. I thought it will represent the whole conversation.