Modelling Residential-Scale Consumer Demographics using Monthly Electricity Consumption Data

Electricity is one of the most widely used forms of energy that plays a significant part in sufficing the fundamental energy demand based on contemporary human needs. It is becoming highly tedious for the energy sector to manage and surveil the modern energy demands based on the constantly changing consumer demographics. In order to progress as a business, it has become pertinent for the distribution companies to evolve their development plans, tariffs, and business models according to the consumer requirements. This article proposes three different models to predict the attributes of a consumer household, namely the multivariate linear regression (MLR), the support vector regression (SVR), and the artificial neural network (ANN). The study uses the PRECON dataset, which is based on the monthly electricity consumption of households in Lahore, Pakistan. All of the proposed models play significant roles in predicting the required consumer demographics for forecasting. The linear model shows the ability to predict the number of people with very low MAPE of 3.57% as compared to other models. So far, ANN has shown the best results in predicting the number of fans, air conditioners, and rooms. However, the MAPE reports extracted from this study show the inability of the used models to explain the variation of property area confidently.