Objectives: This study aims to forecast the daily peak electricity load in Jordan using a dataset of hourly peak load data for the period from January 1, 2010, to December 31, 2022, compiled by the National Electric Power Company (NEPCO). Methods: This study employs the Seasonal Autoregressive Integrated Moving Average (SARIMA) model to make forecasts. The data exhibits an upward trend, seasonality, and non-constant variance. To address these features, the SARIMA model is used to account for the trend and seasonality, while a Box-Cox transformation is applied to manage the non-constant variance. Results: Following the standard Box-Jenkins methodology (identification, estimation, diagnostic checking, and forecasting) and utilizing the “(auto.arima)” function in the RStudio software package, the resulting SARIMA model is ARIMA(1,0,1)(2,1,2)[7]. This model is used to forecast 7 future values of the electricity load. The Mean Absolute Percentage Error (MAPE) and the Root Mean Square Error (RMSE) values, as measures of forecast accuracy, support the precision of our forecasts. Conclusion: Based on empirical results, electricity companies in Jordan are encouraged to use time series models for forecasting electricity loads instead of relying on simple spreadsheet models.
Read full abstract