Reputation: 499
I have a probably very simple question. I have a dataset with rows of different time periods. Every year a new time period is added to the dataset. In my dataset I have period 52, 53 and 54. Soon period 55 will be added. In the code below I extract the rows from the latest period (54):
data new;
set old;
if period=54;
run;
Instead of hardcoding 54 (and soon 55) I want SAS to be able to identify the largest time period, so I don't have to correct the number every time a new period is added to the dataset. I have tried this:
data new;
set old;
if period=max(period);
run;
But the max function requires at least two arguments, and that is not what I want. Can anybody tell me which function to use instead?
Thanks a lot.
Kind regards Maria
Upvotes: 0
Views: 276
Reputation: 9109
You should consider keeping the last period in a data set so you don't have to calculate it each time you need to access the data. Also create an INDEX on period to optimize retrieval.
25 options msglevel=i;
26 data example(index=(period)) lastperiod(keep=period);
27 set sashelp.class(in=in1) sashelp.class(in=in2) end=eof;
28 period = ifn(in1,54,55);
29 output example;
30 if eof then output lastperiod;
31 run;
32
33 data _null_;
34 set period;
35 call symputx('LastPeriod',period);
36 run;
37 data period;
38 set example;
39 where period eq &lastperiod;
INFO: Index period selected for WHERE clause optimization.
40 run;
Upvotes: 2
Reputation: 415
Creating a subset of old containing only the newest period can be done easily using PROC SQL:
PROC SQL NOPRINT;
CREATE TABLE new AS
SELECT *
FROM old
WHERE period = (SELECT MAX(period) FROM old);
QUIT;
Upvotes: 0
Reputation: 1666
One option would be to use a macrovariable and a proc sql step to get the maxperiod, e.g.:
proc sql;
select max(period) into: maxp from old;
quit;
data new;
set old;
if period = "&maxp";
run;
Upvotes: 1