Problem: Some of the premises for a transformer have less than 12 monthly readings (1 year). In order to 'fill in' the missing readings for a premise, I need to do the following: Get the sum of the readings for a premise lacking all 12: This is known as variable Cumuse. Find the other premises for that transformer, locate the readings for the same time period, sum the readings. Determine which premise has the closest summed readings to the cumuse. Using that premise, compute the percent difference between the known sums. Create the missing readings with the closest premise readings times the percent difference.

I am primarily interested in approaches to use for items 2 and 3.

Other information: Numreads tells how many readings there are for a given premise. Totaluse is the actual reading for the Readdate. Readdate is an actual SAS date, not just the 1st day of the month. Readdates cross years. Currently dates are for 1999 and 2000. More than one premise for a given transformer may need readings filled in. I am working with a subset of the data to develop an approach. Once developed, the approach will be used for all transformers. Can process with 6.12 or 8.1 on MVS (OS390).

Here is a partial print of the data available: (See attached file: print)

Thank you in advance for any help you may be able to provide.

______________________________________________________ C. Royce Claytor Dominion Resources Services, Inc. EDI Process Group Phone:804-775-5552 Email:Royce_Claytor@dom.com