Modeling of the energy demand of the residential sector in the United States using regression models and artificial neural networks

This paper describes the development of energy-demand models which are able to predict the future energy demand in the residential sector of the United States. One set of models use an artificial neural network (ANN) technique, and the other set of models use a multiple linear regression (MLR) technique. The models are used to forecast future household energy demand considering different scenarios for the growth rates of the effective factors in the models. The household sector includes all energy-consuming activities in residential units (both apartments and houses) including space and water heating, cooling, lighting and the use of appliances. In order to understand the evolution of household energy use, a set of indicators has been developed. For instance, several factors affect energy consumption for space heating as a share of households' energy demand. These factors include, dwelling size, number of occupants, the efficiency of heating equipment and the useful energy intensity. The paper also analyzes the trend of energy consumption in the residential sector of the United States. Moreover, the effects of important indicators on the energy consumption are discussed. The analysis performed in this paper is done for each census region, where possible, to elucidate the effects of different indicators in each region.