Multiple Output Statements within a Single Data Step

[kisumsam]

Hi there, I have a question about multiple output statements within a data step:

data TWO;
input x y;
datalines;
5 2
3 1
5 6
;
RUN;

data ONE TWO OTHER;
set TWO;
if X eq 5 then output ONE;
if Y lt 5 then output TWO;
output;
run;

The code above seems fairly straightforward. However, when I'm looking at the ONE data set, the order of the observations seems to be a little strange:

From the first output statement, the first two observations should be:

5 2

5 6

From the last output statement, the last three observations should be:

5 2
3 1
5 6

I am expecting the combined data set to be:

5 2
5 6

5 2
3 1
5 6

However, the order of the output data set is:

5	2
5	2
3	1
5	6
5	6

The order of the data set is changed but it wasn't exactly sorted by X and Y.

Does anyone know how the data set is sorted when we have the two output statements in the data step?

[PeterClemmensen]

First the data step reads the observation 5 2. Then it checks if X eq 5, which is true, so it outputs to the data set ONE. next it checks if Y lt 5, which is false, so nothing is outputted. Next you simply have an output statement, which means that the 5 2 observations is outputted again(to all created data sets, including X). That is why the first two observations are 5 2.

[ballardw]

---

@kisumsam wrote:

Hi there, I have a question about multiple output statements within a data step:

data TWO;
input x y;
datalines;
5 2
3 1
5 6
;
RUN;

data ONE TWO OTHER;
set TWO;
if X eq 5 then output ONE;
if Y lt 5 then output TWO;
output;
run;

The code above seems fairly straightforward. However, when I'm looking at the ONE data set, the order of the observations seems to be a little strange:

I am expecting the combined data set to be:

5 2
5 6

5 2
3 1
5 6

Does anyone know how the data set is sorted when we have the two output statements in the data step?

---

To get that specific result one way would be:

data three;
   set TWO (where=(x ge 5))
       two
   ;
run;

which reads set two twice, the first bit says to keep only the records with x ge 5 and since that is the first input set they are output first,

then read set two again with all records.