Date:         Wed, 8 Oct 1997 15:10:42 -0700
Reply-To:     "M. Fang" <fangm@UVIC.CA>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         "M. Fang" <fangm@UVIC.CA>
Subject:      identify new ids
Content-Type: TEXT/PLAIN; charset=US-ASCII

Thank you very much for all of you who responded to my question. However,
I am terribly sorry that I did not write out my question clearly. Let me
try again.
 
In the dataset, there is one variable named 'ID', and there is another
variable named 'Time'. Time starts from 199608 to now, and ID records
whoever registers in a given month. Each person is given an unique ID. Now
we are trying to create a monthly report to show new ids joined in for the
month and old ids left for the month (that is people who left the
program) over this period. My question is how do I identify the new ids in
a given month, and how to find out who left (that is, for example, to find
out the ids appeared in Feb, but disappeared in Mar). One important
point, ids which disappeared may appear again. The data itself is
rolled-up. That is new ids (include those who come back) will be
added to the dataset. The original data looks like this:
 
ID    Time
1234 199612
2345 199612
1234 199701
2345 199701
3456 199701
4567 199701
1234 199702
2345 199702
3456 199702
6789 199702
7890 199702
1234 199703
2345 199703
3456 199703
6789 199703
7890 199703
4567 199703
 
So, for example, '4567' in Jan is not present in Feb, but it is back in
Mar, and '6789' and '7890' are present in Feb (new users). In the count
for new users, we only want to include first-timers, not those comebacks.
 
I hope I have presented my question better this time. Again, any help
would be appreciated.
 
 
Minot