| Abstract [eng] |
With the implementation of the European Green Deal in EU member states, including Lithuania, consumption of renewable energy is on the rise. Such shifts fundamentally transform the structure of national energy sectors, shape the priorities of legislative and executive authorities, create new opportunities for member states and alter their relative advantages. For Lithuania, this initiative also presents the opportunity to become a strategically important country, a net exporter of green energy. Therefore, it is crucial to understand the overall trend in the demand for renewable energy resources within the country and how it will evolve in the upcoming years. Future trends are forecasted using various mathematical methods: time-series models, econometric models, ARIMA-family models, grey models and deep-learning models. External cross-validation is employed to assess model accuracy, while hyperparameters are tuned via internal cross-validation. Models within each family are selected by experimenting with different alternatives and modifications according to uniform error metrics (MAPE, MAE, RMSE). The exponential grey model EXGM (1,1) proved to be the best model for forecasting the available dataset. |
