CODEWITHSUNDEEP
sas
Beth
Beth

Reputation: 55

SAS new variable based on new variable's preivous value

I have a time series data set of over 3,000 observations. I am trying to create a new variable that is dependent on that variable's previous value and I am having trouble getting SAS to calculate each observation in order to ensure that the previous value is calculated before it gets to the next observation.

I have set the first observation to 1. Then for the remaining observations I want SAS to calculate:

New_var = lag(new_var) * (1 + Var2)

Any ideas? I am sure it is a relatively easy answer, but I haven't been able to decipher it myself.

Upvotes: 1

Views: 124

Answers (2)

MrCurious
MrCurious

Reputation: 333

I had the same question and reached here searching for it. I came out with the following logic and so thought of answering it here as well for future SAS enthusiasts:

  1. Use (or create) an existing (or new) variable which has unique values. You may use retain statement in SAS or monotonic() function in proc sql as shown below:

    /using proc sql/ proc sql; create table temp6 as select a.*, monotonic() as var3 from temp5 as a; quit;

    /using data step/ data temp7; set temp5; retain var3; var3 + 1; run;

  2. Once you have this variable, you may use the retain statement and first. flag to calculate New_Var as below. I have initialized the New_Var with 1 for the very first row. You may change the initial value accordingly.

    data temp8; set temp7; by var3; retain New_Var 1; if first.var3 then New_Var = New_Var * (1 + Var2); run;

Hope this helps!!

Upvotes: 0

DanZ
DanZ

Reputation: 481

You could try a retain statement instead of lag.

retain new_var;
new_var = new_var * (1+Var2);

Upvotes: 3

Related Questions