Data-Efficient Electricity Price Forecasting Using Event-Based Market Representations
| dc.contributor | Mateo Fornés, Jordi | |
| dc.contributor | Universitat de Lleida. Escola Politècnica Superior | |
| dc.contributor.author | Fernández Sisquella, Hugo | |
| dc.date.accessioned | 2026-09-02T06:55:19Z | |
| dc.date.available | 2026-09-02T06:55:19Z | |
| dc.date.issued | 2026-07 | |
| dc.description.abstract | Electricity price forecasting is a challenging task due to the volatility of electricity markets and the large amount of data required to represent their behaviour. This work studies onehour-ahead electricity price forecasting from the perspective of data representation, focusing on whether event-based representations can reduce the amount of market data used for training while preserving predictive performance. A Nord Pool day-ahead market dataset is constructed using prices, buy and sell volumes, cross-border flows and transmission capacities. From this dataset, three input representations are compared: the full raw dataset, an automatic rbaTheta-based event representation and a manually defined market event representation based on price, volume, flow and capacity events. The proposed methodology is evaluated on Norwegian bidding zones with different volatility levels using several forecasting models and RMSE as the main evaluation metric. The results show that market-informed event representations can achieve forecasting performance close to the full dataset while using substantially fewer raw values. This suggests that carefully analysing and selecting relevant market situations can provide a compact and interpretable alternative to using all available market data. | |
| dc.format.extent | 46 p. | |
| dc.identifier.uri | https://hdl.handle.net/10459.1/471122 | |
| dc.language.iso | eng | |
| dc.rights | cc-by-nc-nd | |
| dc.rights | Attribution-NonCommercial-NoDerivatives 4.0 International | * |
| dc.rights.accessRights | info:eu-repo/semantics/openAccess | |
| dc.rights.uri | http://creativecommons.org/licenses/by-nc-nd/4.0/ | * |
| dc.subject | Data-Efficient Electricity | |
| dc.subject | Price Forecasting | |
| dc.subject | Using Event-Based Market Representations | |
| dc.title | Data-Efficient Electricity Price Forecasting Using Event-Based Market Representations | |
| dc.type | info:eu-repo/semantics/bachelorThesis |