SAS aggregate rows computation - sas

Question

I'm a beginner user of SAS especially when it comes to aggregate rows computation.
Here is a question which I believe some of you may have encountered before.
The data I have is related to insurance policies, here is an example dataset: columns from left to right are customer number, policy number, policy status, policy start date and policy cancel date (if the policy is not active, otherwise is a missing value).
data have;
informat cust_id 8. pol_num $10. status $10. start_date can_date DDMMYY10.;
input cust_id pol_num status start_date can_date;
format start_date can_date date9.;
datalines;
110 P110001 Cancelled 04/12/2004 10/10/2013
110 P110002 Active 01/03/2005 .
123 P123001 Cancelled 21/07/1998 23/04/2013
123 P123003 Cancelled 22/10/1987 01/11/2011
133 P133001 Active 19/02/2001 .
133 P133001 Active 20/02/2002 .
;
run;
Basically I want to roll these policy level information to customer level, if a customer holds at least one active policy, then his status would be 'Active', otherwise if all his policies are Cancelled, then his status becomes 'Inactive'. I also need a customer "start date" which picks up the earliest policy start date under that customer. If the customer is 'Inactive', then I need the customer's latest policy cancel date as the customer's exit date.
Below is the what I needed:
data want;
informat cust_id 8. status $10. start_date exit_date DDMMYY10.;
input cust_id status start_date exit_date;
format start_date exit_date date9.;
datalines;
110 Active 01/03/2005 .
123 Inactive 22/10/1987 23/04/2013
133 Active 19/02/2001 .
;
run;
Solution in any form would be much appreciated! Either DATA step or PROC SQL is fine.
Thank you so much.

Answer 1

You can do something like that:
proc sql;
create table want as
select cust_id,
case when count(case when status='Active' then 1 end) > 0
then 'Active'
else 'Inactive'
end as status,
min(start_date) as start_date,
case when count(case when status='Active' then 1 end) = 0
then max(can_date)
end as exit_date
from have
group by cust_id;
quit;

Answer 2

You could attack the question in a DATA step. Here's one simple way, assuming your data are sorted by cust_id and start_date...
data want;
set have (keep=cust_id status start_date exit_date);
where upcase(status) contains 'ACTIVE';
by cust_id start_date;
if first.start_date then output;
else delete;
run;
/*BEGINNER NOTES*/
*1] WHERE tells SAS to compile only records that fit a certain
condition - the DS 'want' will never have any observations with
'CANCELLED' in the status variable;
*2] I use UPCASE() to standardize the contents of status, as CONTAINS
is a case-sensitive operator;
*3] FIRST.variable = 1 if the value is the first encountered in
the compile phase;