Comparative Performance of ARMA and ARIMAX Models in Time Series: A Case Study of Urban Electricity Consumption of Sulaymaniyah, Iraq

Abstract

Bridging the gap between supply and demand can be greatly assisted through modeling and forecasting daily energy consumption. The aim of this study is to utilize the Autoregressive Integrated Moving Average (ARIMA), and Autoregressive Integrated Moving Average with Exogenous Variables (ARIMAX), models to forecast and modelling energy consumption in Sulaymaniyah province, given the critical nature of this issue. The aforementioned models were developed employing daily electricity consumption and meteorological data from January 1, 2024, to December 31, 2024. The Akaike Information criteria (AIC) and Bayesian Information criteria (BIC) have been utilized in Google Colab software to determine the appropriate values of the parameters of the described models (p, d, q). The most effective models were determined to be ARIMA (3,0,4) and ARIMAX (3,0,3). It has been concluded from the comparison of the Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Coefficient of Determination values (R-Square) for the models. the ARIMA model with a (MAE = 0.0470), (RMSE = 0.060) and (R-Square = 70%) was more precise than the ARIMAX model in modeling and forecasting power consumption.