Retrieve a record with max time when both records are same

[chennupriya]

datea    time                     customer
2015-11-20  11:18:36.0000000  lucy
2015-11-20  11:21:09.0000000         lucy

dateb      amount
2015-11-20  250

output

datea    time                     customer  amount
2015-11-20  11:18:36.0000000  lucy   250
2015-11-20  11:21:09.0000000         lucy   250

I have three tables and i have an output where it has two records but the amounts are duplicated so how to pick a record with max time . can anyone pls help

[dcruik]

This should eliminate your duplicates and keep the record with the max time:

data output;
input datea time customer$ amount;
informat datea yymmdd10. time time10. customer $10. amount 6.;
format datea yymmdd10. time time10. customer $10. amount 6.;
datalines;
2015-11-20 11:18:36 lucy 250
2015-11-20 11:21:09 lucy 250
;
run;

proc sort data=output;
by datea customer descending time;
run;

data want;
set output;
by datea customer descending time;
If First.customer;
run;

[FreelanceReinh]

You pointed out that the amounts are duplicated. So, would you want to keep both records if the amounts were different, say, 240 and 250?

How would your data look like if there had been two completely different transactions which happened to involve the same amount? For example, Lucy paid 250 for a pair of shoes and another 250 for a coat on the same day. Would it make sense to pick only one of these transactions?

Would you still want to pick only one of the two records if they were, say, 12 hours apart rather than only a few minutes?

What if the first datetime was 2015-11-20 23:58:36 and the second was 2015-11-21 00:01:09? In this case, grouping by datea would separate them and both would be selected.

[chennupriya]

hi ,

what if the data is like this

datea    time                     customer  amount sssid
2015-11-20  11:18:36.0000000  lucy   250    xxx
2015-11-20  11:21:09.0000000         lucy   250    xxx
2015-11-23  11:20:35.0000000            lucy   500    xxx

and output should be

datea    time                     customer  amount sssid
2015-11-20  11:21:09.0000000         lucy   250    xxx
2015-11-23  11:20:35.0000000            lucy   500    xxx

[Reeza]

The concept is correct, use a sort and BY group processing. Sort such that for duplicate records the latest time is at the top and then take the first record for each group.

proc sort data=output;
by datea amount descending time;
run;

data want;
set output;
by datea amount;
if first.amount;
run;

[PGStats]

If everything is the same except the time in those duplicates, then you can simply do:

proc sql;
create table want as
select datea, max(time) as time, customer, amount, sssid
from have
group by datea, customer, amount, sssid;
quit; 

[dcruik]

If you have multiple datea values, then I would just modify the sort by variables and what you are using as the variable with your First. satement.  This would get you what you want with the updated data you provided:

data have;
input datea time customer$ amount sssid$;
informat datea yymmdd10. time time20. customer $10. amount 6. sssid $10.;
format datea yymmdd10. time time20. customer $10. amount 6. sssid $10.;
datalines;
2015-11-20 11:18:36 lucy 250 xxx
2015-11-20 11:21:09 lucy 250 xxx
2015-11-23 11:20:35 lucy 500 xxx
;
run;

proc sort data=have;
by customer datea descending time;
run;

data want;
set have;
by customer datea descending time;
If First.datea;
run;