regression with dates

[deleted_user]

I am working on a regression model that is trying to estimate the energy demand across the year.

A previous attempt at this model used Heating Degree Days and month (actually an indicator of month of the year that is handed to a stepwise regression).

I am hoping to improve on the model. There are obvious problems with repeated measures etc in the data.

I'm wondering if there is a regression I can do in SAS using longitudinal data with repeated measures and if anyone can suggest a reference?

[Doc_Duke]

look at cyclic variability models. You transform the dates into something like degrees on the compass and then do the regression. It allows you to look at various sinusoidal type patterns. It helps to have a postulated model ( days of the week, say). I'm rusty on it, but there was a good bit of literature on it.