Select only the first observation within a string of similar observations

[apolitical]

In the above table, I now have variables Account, Day and Alert, and would like to obtain Alert_Real. The logic is that for each account, if alerts appear on consecutive days, only the first day of that string of successive alerts is marked in Alert_real, whereas the following days are recorded as no alert. Such cases are highlighted in read. And I would like to do so for each account. Hope I made myself clear. Thanks a lot.

[Astounding · Accepted solution]

Assuming your data set is sorted by ACCOUNT:

data want;

set have;

by account alert notsorted;

if first.alert then alert_real = alert;

else alert_real=0;

run;

[apolitical]

Thanks, that takes care of it. The by statement with "notsorted" was what's needed.