CODEWITHSUNDEEP
databaseloopsif-statementmacrossas
Roberto
Roberto

Reputation: 729

Macro loop in SAS - passing value to condition

I'm a beginner in SAS and I am struggling a bit with the macro loop in SAS. The problem is illustrated by the code below. The task here is to create separate subsets and save them as libraries for later post-processing. Additionally I added graphs for visualization. I am operating on a huge database but for this post I create a sample at the beginning of the code for simplification.

However, it seems that the internal condition (IF ID = i ) doesn't filter out the data. Instead the internal loop creates empty tables (but with correct names: "SUB1", "SUB2", "SUB3") with a column (variabale) called "i".

DATA EXAMPLE;

INPUT ID DATE DDMMYY8. VALUE;
FORMAT DATE DDMMYY8.;
DATALINES;
1 01012011 100
1 01022011 400
1 01032011 678
2 01012011 678
2 01022011 333
2 01032011 333
3 01012011 733
3 01022011 899
3 01032011 999
;
%MACRO filter(number);
    %DO i=1 %TO &number;
        DATA SUB&i;
        SET WORK.EXAMPLE;
        IF ID = i;
        PROC SGPLOT DATA=SUB&i;
        reg x=DATE y=VALUE;
        RUN;
    %END;
%mend filter;

%filter(3);

If I manually copy and paste the part inside macro and manually change i to numbers 1 to 3 it creates correct graphs. What is wrong in this code? How can I pass the value from the DO statement inside the code?

I am using SAS Studio.

Upvotes: 0

Views: 743

Answers (3)

user9166085
user9166085

Reputation: 33

One tip for debugging your macro code is to add the statement

options mprint;

This will show the code that SAS is actually using.

For example in the log:

70   options mprint;
71   %MACRO filter(number);
72       %DO i=1 %TO &number;
73           DATA SUB&i;
74           SET WORK.EXAMPLE;
75           IF ID = &i;
76           PROC SGPLOT DATA=SUB&i;
77           reg x=DATE y=VALUE;
78           RUN;
79       %END;
80   %mend filter;
81
82   %filter(2);
MPRINT(FILTER):   DATA SUB1;
MPRINT(FILTER):   SET WORK.EXAMPLE;
MPRINT(FILTER):   IF ID = 1;

NOTE: There were 9 observations read from the data set WORK.EXAMPLE.
NOTE: The data set WORK.SUB1 has 3 observations and 3 variables.
NOTE: DATA statement used (Total process time):
      real time           0.03 seconds
      cpu time            0.01 seconds


MPRINT(FILTER):   PROC SGPLOT DATA=SUB1;
MPRINT(FILTER):   reg x=DATE y=VALUE;
MPRINT(FILTER):   RUN;

Upvotes: 0

Richard
Richard

Reputation: 27526

The macro is creating empty data sets because the code that the macro eventually writes contains the subsetting if statement

if ID = i;

Because the data set does not contain a variable i a new variable named i is added to the PDV and the output data sets SUB1, SUB2, SUB3. The default value for i is missing and no ID value is missing, thus no rows pass the test and you get empty data sets. The log will also provide clues to the situation:

NOTE: Variable i is uninitialized.

When abstracting a code segment for 'macroization' be sure to use & in front of the macro variables. Thus, when the macro contains

if ID = &i;

The eventual code written by the macro system will have your 3 similar code operations with the different values of the macro variable.

...
if ID = 1;
...
...
if ID = 2;
...
...
if ID = 3;
...

Upvotes: 2

Tom
Tom

Reputation: 51621

Right now you are producing the same graph three times because the datasets SUB1, SUB2, SUB3 all use the same set of data. That is because the only thing in your data step that depends on the value of the macro variable I is the name.

You are currently selecting the observations where the variable ID matches the variable I. Perhaps you meant to select the observations where the variable ID matches the macro variable used in the %DO loop?

IF ID = &i;

Upvotes: 0

Related Questions