The way I solved this was to use proc sql. I added a URN to each record using _N_ in a datastep and then joined on urn=urn-1.    Thanks for the other replies.   Regards   Andrew ... View more

It depends a bit on your data structure  - quite a bit - and how your'e defining month. If months are 1 to 12 are the same for each customer for example, or do they relate to when a customer first joined the company.    Also you may want to use survival analysis instead of just times as that accounts for accounts still open/active more accurately than just just using accounts.    @andrewjmdata wrote: Hi, I have a table of 100,000 customers. I have the month when they started buying from my client. The range of months is 1-200.   What I want to do is break that down into 1/3rd's for example out put might be:   1st 1/3 (33%) of customers bought in months 1->12  2nd 1/3 (33%) of customers bought in months 13->124 3rd 1/3 (33%) of customers bought in months 125->200   I think the solution is to use proc rank but can't seem to get month breaks based on volume of customers.   Make sense?   Thanks   Andrew      ... View more

That doesn't matter. Anything on the commandline is visible to all processes and all users. ... View more