topic Re: difference between two calculation in SAS Programming

difference between two calculation

ttxq — Tue, 06 Jun 2017 19:00:40 GMT

Hi all,

I want to know what is the differences between these two statements: (highlighted in red color)

proc sort data=com;
by dis;
run;

data com;
set com;
by dis;

if first.dis then disn+1; (if first.dis then disn=disn+1)
run;

when I run each program, the value of disn is different. I want to know the difference between disn+1 and disn=disn+1.

Thanks

Re: difference between two calculation

WarrenKuhfeld — Tue, 06 Jun 2017 19:12:54 GMT

dsn + 1;

is a sum statement. It implicitly adds:

retain dsn 0;

dsn = dsn + 1;

is an assignment statement. There is no implicit retain.

Re: difference between two calculation

ballardw — Tue, 06 Jun 2017 19:16:52 GMT

When you use a statement like this:

if first.dis then disn+1;

Then the variable DISN is created with an implied Retain statement which means the value is carried from record to record in the data step. If there is a variable in the dataset name in a SET statement then that value gets reset each time a record is pulled from the dataset.

So the statement as used will increase a count by 1 for each first level of the variable DIS when the statement executes.

if first.dis then disn=disn+1;

If the variable DISN is not declared with a RETAIN statement and this is the first use then the value of Disn starts as missing. When a missing value has 1 added this way the result is missing. Unless there is a variable Disn in your com set I would expect the result to be missing from the addition. If you do have a Disn variable then the result would be the addition of 1 to that value when not missing.

If you need to sum values when something may be missing use the SUM function. However without a retain such as your example the result of disn = sum(disn,1); would likely be 1 for every record.

Re: difference between two calculation

Reeza — Tue, 06 Jun 2017 19:16:56 GMT

To make it the same, add the RETAIN, if desired.

This should make disn and disn2 the exact same.

data com1;
set com;
by dis;
retain disn;

if first.dis then disn=disn+1;

if first.dis then disn2+1;

run;

Re: difference between two calculation

ttxq — Tue, 06 Jun 2017 19:20:20 GMT

I see. Thank you

Re: difference between two calculation

ttxq — Tue, 06 Jun 2017 19:20:38 GMT

Thanks a lot