Statistical energy savings calculations are fundamentally rooted in how well energy data can be normalized against influencing factors. Attempts to predict monthly energy use in academic buildings based strictly on weather as a driver for energy fail because of variable monthly occupancy. A genetic based energy model is used to characterize monthly energy consumption in academic buildings or any other buildings with variable occupancy. Such a model is essential for both estimating savings when changes are made and for continuously commissioning the building. Monthly average outdoor air temperature is considered to reflect the weather driver on energy use. Monthly occupancy is based upon the the historical academic year calendar; occupancy is considered a linear function of number of normal academic days per month.. The multi-functional model developed is tested on both simulated and actual academic building energy data. The results demonstrate universally improved correlations.