Modeling residential electricity demand using neural network and econometrics approaches

Electricity industry is one of the main foundations of each country. Electricity demand growth in developing countries is at its peak so that these countries play an important role in electricity consumption. Since this energy can´t be stored in large quantity, forecasting consuming load is a major concern in using electricity energy. This paper has formulated electrify consumption for residential sector using neural network and econometrics models. The first step is determining effective variables on electricity consumption. These variables are electricity price, gas price and the amount of previous year consumption. The data related to monthly electricity consumption was collected from the Iran´s statistics center. To serve this purpose, neural network and econometrics models have been used and the results of these two approaches have been compared. The results propose using log-linear models. The results also show that there is a direct relationship between the electricity consumption and its amount the previous year. Also gas price and electricity consumption has a direct relationship, too. Electricity consumption and its price have an indirect relationship which proves negative demand slope. A comparison between two named approaches concludes that the neural network is more capable in evaluating the actual consumption model.