Modeling of gas consumption in the city

Abstract

Based on the data collected over a two year time period, which included temperature, wind speed and gas consumption during the day, the effects of weather factors on gas consumption in the city have been established with the use of multiple regression. The impact of a particular month, day (dummy variable) or holiday of a year on the gas consumption has also been determined. The models of linear regression and artificial neural networks have been constructed for determining the gas consumption. An attempt has been made to find the best regression models and compare them to the neural network models with the use of mean absolute percentage error (MAPE).