Forecasting Swedish FCR-D Prices using Penalized Multivariate Time Series Techniques
(2023) DABN01 20231Department of Economics
Department of Statistics
- Abstract
- The Swedish energy market is becoming more and more sustainable, with an increasing volume and number of diversified energy sources being continuously added to the mix. To stabilize the grid frequency, auctions are held to offer energy providers incentives to produce or consume energy on short notice. This paper is applying different multivariate time series models and their ensemble to find reliable forecasts. Given the high dimensionality through a variety of factors influencing the energy market, penalized models are used to perform variable selection and obtain sparser models. The investigated data contains a lot of noise. Therefore, part of the work focuses on the effect of noise filtering. The goal is to create reliable price... (More)
- The Swedish energy market is becoming more and more sustainable, with an increasing volume and number of diversified energy sources being continuously added to the mix. To stabilize the grid frequency, auctions are held to offer energy providers incentives to produce or consume energy on short notice. This paper is applying different multivariate time series models and their ensemble to find reliable forecasts. Given the high dimensionality through a variety of factors influencing the energy market, penalized models are used to perform variable selection and obtain sparser models. The investigated data contains a lot of noise. Therefore, part of the work focuses on the effect of noise filtering. The goal is to create reliable price forecasts which may help sustainable energy providers maintain their position or enter this market, stabilize the grid, and help Sweden make the transition to renewable energy. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9133997
- author
- Wunderlich, Franz Lennart LU and Brugger, Sebastian LU
- supervisor
- organization
- course
- DABN01 20231
- year
- 2023
- type
- H1 - Master's Degree (One Year)
- subject
- keywords
- Swedish Energy Market, Multivariate Time-Series, Lasso, Forecasting, Noise Filtering
- language
- English
- id
- 9133997
- date added to LUP
- 2023-11-21 12:54:59
- date last changed
- 2023-11-21 12:54:59
@misc{9133997, abstract = {{The Swedish energy market is becoming more and more sustainable, with an increasing volume and number of diversified energy sources being continuously added to the mix. To stabilize the grid frequency, auctions are held to offer energy providers incentives to produce or consume energy on short notice. This paper is applying different multivariate time series models and their ensemble to find reliable forecasts. Given the high dimensionality through a variety of factors influencing the energy market, penalized models are used to perform variable selection and obtain sparser models. The investigated data contains a lot of noise. Therefore, part of the work focuses on the effect of noise filtering. The goal is to create reliable price forecasts which may help sustainable energy providers maintain their position or enter this market, stabilize the grid, and help Sweden make the transition to renewable energy.}}, author = {{Wunderlich, Franz Lennart and Brugger, Sebastian}}, language = {{eng}}, note = {{Student Paper}}, title = {{Forecasting Swedish FCR-D Prices using Penalized Multivariate Time Series Techniques}}, year = {{2023}}, }