Data-driven forecasting of Finnish electricity prices
Balashova, Aleksandra (2025)
2025
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025121737539
https://urn.fi/URN:NBN:fi:amk-2025121737539
Tiivistelmä
This thesis investigates whether electricity prices in Finland can be accurately predicted using time series and machine learning techniques. Electricity prices fluctuate daily due to various factors such as weather conditions, demand levels, and generation capacity, making it challenging for consumers and companies to plan their energy costs.
The main objective was to evaluate the accuracy of hour-ahead electricity price forecasts and identify which forecasting methods perform best for this purpose. The study also sought to determine how much historical data is required to achieve reliable short-term predictions.
For the study, data on electricity prices in Finland over the past 10 years was collected from an open API. The dataset was cleaned and validated before being used for model training and evaluation. Subsequently, multiple forecasting methods were compared based on established performance metrics.
The outcomes of this thesis comprise a validated dataset, an evaluation of forecasting models, and insights into prediction accuracy and limitations. The findings help to understand how data-driven methods can improve electricity price forecasting and support informed energy planning in Finland.
The main objective was to evaluate the accuracy of hour-ahead electricity price forecasts and identify which forecasting methods perform best for this purpose. The study also sought to determine how much historical data is required to achieve reliable short-term predictions.
For the study, data on electricity prices in Finland over the past 10 years was collected from an open API. The dataset was cleaned and validated before being used for model training and evaluation. Subsequently, multiple forecasting methods were compared based on established performance metrics.
The outcomes of this thesis comprise a validated dataset, an evaluation of forecasting models, and insights into prediction accuracy and limitations. The findings help to understand how data-driven methods can improve electricity price forecasting and support informed energy planning in Finland.