What is the difference between select when and if ,else if in sas datastep can any one help

[sas_Forum]

What is the difference between select when and if ,else if in sas datastep can any one help

[Ksharp]

In my eyes, there is no difference between them.

But I would use select when you have lots of conditions to judge, because it is look elegant and succinct.

Ksharp

[ballardw]

I would say the major difference is that If , else if allows more use of compound comparisons that may miss all of the cases you need to consider.

As Ksharp says if I need to consider more than 2 or 3 values for a single variable Select is easier to understand and I believe is supposed to execute quicker as it branches differently instead of evaluating a whole series of If then else if ....

The OTHERWISE statement with SELECT also serves as a better catch-all than a last "else" because different nesting in a series may not be as obvious that none of the criteria had been met.

I think the most cases I have dealt with in a single select when is around 194. I am not sure if you could even nest if then else that deep.

[AjayKant]

Just the way you use both work in similar phase and no need to understand difference because it give you same result .

as per your comfort you can either choose if then else or select when statement ;

I tried it on same dataset and find out the difference is how much they are used memory .

Dataset of 15000 observation and 20 variables is used to analyse .

Codes  processed using IF

data if  ;

set raw ;

if Region='Central' then do ;

code=1;

end ;

if Region='East' then do ;

code=2;

end ;

if Region='Wast' then do ;

code=3;

end ;

else code=0 ;

run ;

*----------------------------------------*

NOTE: There were 15000 observations read from the data set WORK.RAW.

NOTE: The data set WORK.IF has 15000 observations and 18 variables.

NOTE: DATA statement used (Total process time):

other is using select when

data when ;

set raw ;

select (Region);

when(central)do ;

code=1 ;

end ;

when(East)do ;

code=2 ;

end ;

when(West)do ;

code=3 ;

end ;

otherwise code=0 ;

end ;

run ;

----------------------------------------------

NOTE: There were 15000 observations read from the data set WORK.RAW.

NOTE: The data set WORK.WHEN has 15000 observations and 21 variables.

NOTE: DATA statement used (Total process time):

Timing used for processing IF

      real time           0.03 seconds

      user cpu time       0.01 seconds

      system cpu time     0.01 seconds

      memory              362.75k

      OS Memory           6584.00k

      Timestamp           01/17/2014 07:48:49 PM

Timing used for processing select when ;

      real time           0.09 seconds

      user cpu time       0.01 seconds

      system cpu time     0.09 seconds

      memory              367.62k

      OS Memory           6584.00k

      Timestamp           01/17/2014 07:51:27 PM

You find a bit difference in real time and cpu time ..

here is start using the way if data will increase to millions which SAS system  actual have to work .

the difference can me measured easily .

I hope it will help you understand .

[ballardw]

Try the time difference coding

if Region='Central' then

Else if Region='West' then

else if Region='East' then

else

[AjayKant]

Can you explain what is time difference in content of this....

[CTorres]

You can simplify the select statement eliminating the do-end blocks innecesary in this case:

data when ;

set raw ;

select (Region);

when('Central') code=1 ;

when('East') code=2 ;

when('West') code=3 ;

otherwise code=0 ;

end ;

run ;

[AjayKant]

Thanks and its reduce the slight tie as well but yes in terms of performance if function works well .

The stats i get after re-edit the codes .

NOTE: There were 15000 observations read from the data set WORK.RAW.

NOTE: The data set WORK.IF has 15000 observations and 18 variables.

NOTE: DATA statement used (Total process time):

      real time           0.03 seconds

      user cpu time       0.00 seconds

      system cpu time     0.01 seconds

      memory              363.84k

      OS Memory           6584.00k

      Timestamp           01/17/2014 08:47:50 PM

NOTE: There were 15000 observations read from the data set WORK.RAW.

NOTE: The data set WORK.WHEN has 15000 observations and 21 variables.

NOTE: DATA statement used (Total process time):

      real time           0.12 seconds

      user cpu time       0.03 seconds

      system cpu time     0.09 seconds

      memory              365.62k

      OS Memory           6584.00k

      Timestamp           01/17/2014 08:47:50 PM

after re-editing the function works great but a major in there real time and  CPU time and this is the only difference I found .