Compute time intervals for each customer ID

[Geo-]

how to count time interval for each customer?

have:

id	time
103	07MAR2015:09:13:02.00
102	07DEC2015:10:42:19.00
102	07DEC2015:10:43:39.00
101	07MAY2015:09:13:30.00
102	07DEC2015:10:43:51.00
101	07MAY2015:09:13:41.00
103	07MAR2015:09:13:05.00
101	07MAY2015:09:13:55.00

want:

id	time	interval
101	07MAY15:09:13:30	.
101	07MAY15:09:13:41	11
101	07MAY15:09:13:55	14
102	07DEC15:10:42:19	.
102	07DEC15:10:43:39	80
102	07DEC15:10:43:51	12
103	07MAR15:09:13:02	.
103	07MAR15:09:13:05	3

[VDD]

How does interval get assigned?  your sample reflects that interval is char based on the sample being left justified.

[Geo-]

I have updated and add the seconds.Would it be more clear?thanks

[mkeintz]

You appear to want to do 2 things:

1. Take unsorted data, and order it by ID/TIME.  Is that correct?
   
   
2. There are multiple obs with ID101 and tied time values.  Are there other variables that would force a particular ordering of those tied records?  After all you have interval values of 11 and 14 for the 2nd and 3rd records.  I see no data that would produce those values.
3. Calculate a new variable called interval, which is left missing for the first obs of each id, but has a positive value for all subsequent obs for each id.  As @VDD asked, what it the formula for generating the new variable interval?

[VDD · Accepted solution]

Yes having the seconds included makes it clear how the value is to be determined.

This problem has been solve by PG Stats at this link.  You might have to do a little work with the code so that the variables reference your case information.

https://communities.sas.com/t5/SAS-Programming/count-average-time-interval/m-p/487616#M127080