Predicting energy usage using historical data and linear models

Abstract

This paper presents a method to predict energy usage, based on weather conditions and occupancy, using a multiple linear regression model (MLR) in research office buidings. in this study, linear regression models of four research office sites in different regions of New Zealand were selected to show the capability of simple models to reduce margins of error in energy auditing projects. The final linear regression models developed were based on monthly outside temperatures and numbers of full time employees (FTE's). Comparing actual and predicted energy usage showed that the models can predict energy usage within acceptable errors. The results showed that each building should be investigated as an individual unit.