Accurate prediction of future events is of great interest in various contexts. This thesis focuses on forecasting and predicting energy usage in the industrial sector, which provides valuable information for government agencies to plan and allocate the available budget. More specifically, the purpose is to evaluate if using explanatory variables in a dynamic regression with seasonal autoregressive integrated moving average (SARIMA) errors improves the forecasting accuracy of quarterly energy usage in the industrial sector in Sweden compared to a standard SARIMA model. The SARIMA model used for comparison is SARIMA(1,0,0)(0,1,1), while the dynamic regression model used has the explanatory variable value added of the industrial sector and SARIMA(1,0,0)(0,1,1) errors. The forecast performance of the two models is compared for both quarterly and yearly forecast horizons using root mean squared error (RMSE) and mean absolute error (MAE). The results show that the RMSE and MAE of the dynamic regression model are lower for both forecast horizons compared to the SARIMA model. Also, a significance test (OOS-t) and an encompassing test (ENC-NEW) are employed, which show that the difference in forecasting accuracy is statistically significant and that the SARIMA forecast doesn’t encompass the forecast of the dynamic regression model.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:uu-328292 |
| Date | January 2017 |
| Creators | Anners, Carl |
| Publisher | Uppsala universitet, Statistiska institutionen |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0018 seconds