Treffer: Nordpool electricity price forecasting using artificial intelligence algorithms / ; Nordpool elektros kainų prognozavimas naudojant dirbtinio intelekto algoritmus.

The objective of this bachelor's project is to identify the most optimal AI algorithms for predicting Nord Pool electricity prices and to implement techniques that further enhance prediction accuracy. The techniques include extracting relevant features, hyperparameter tuning, and data splits. The models are then trained with the Nord Pool dataset, which includes data such as electricity prices, consumptions, productions, wind productions and cross-border exchanges. This data has been extracted and formatted from Nord Pool servers. The models used in this study include linear regression, K-nearest neighbors regression, support vector regression, extreme gradient boosting, and multivariate long short-term memory models. Each model is fitted with a distinct set of hyperparameter combinations, and the resulting performance is evaluated. Once all models have been trained and evaluated according to the specified metrics including mean squared error (MSE), mean absolute error (MAE), root mean squared error (RMSE), and R², they are compared to one another in order to draw conclusions.