Short-term electricity price forecasting - evaluation of selected hybrid models

• Název práce v češtině: Krátkodobé předpovídání cen elektřiny s užitím hybridních modelů
• Název v anglickém jazyce: Short-term electricity price forecasting - evaluation of selected hybrid models
• Akademický rok vypsání: 2016/2017
• Typ práce: bakalářská práce
• Jazyk práce: angličtina
• Ústav: Institut ekonomických studií (23-IES)
• Vedoucí / školitel: prof. PhDr. Ladislav Krištoufek, Ph.D.
• Řešitel: skrytý - zadáno vedoucím/školitelem
• Datum přihlášení: 09.11.2016
• Datum zadání: 09.11.2016
• Datum a čas obhajoby: 14.06.2017 09:00
• Místo konání obhajoby: Opletalova - Opletalova 26, O105, Opletalova - místn. č. 105
• Datum odevzdání elektronické podoby: 17.05.2017
• Datum proběhlé obhajoby: 14.06.2017
• Oponenti: Mgr. Júlia Jonášová

Seznam odborné literatury

1. Che, Jinxing, and Jianzhou Wang. "Short-term Electricity Prices Forecasting Based on Support Vector Regression and Auto-regressive Integrated Moving Average Modeling." Energy Conversion and Management 51.10 (2010): 1911-917.
2. Alvarez, Francisco Martinez, Alicia Troncoso, Jose C. Riquelme, and Jesus S. Aguilar Ruiz. "Energy Time Series Forecasting Based on Pattern Sequence Similarity." IEEE Transactions on Knowledge and Data Engineering 23.8 (2011): 1230-243.
3. "Day-ahead Electricity Price Forecasting by a New Hybrid Method." Day-ahead Electricity Price Forecasting by a New Hybrid Method. N.p., n.d. Web. 13 Feb. 2017.
4. "Electricity Price Forecasting: A Review of the State-of-the-art with a Look into the Future." Electricity Price Forecasting: A Review of the State-of-the-art with a Look into the Future. N.p., n.d. Web. 13 Feb. 2017.
5. Jin, C. H., Pok, G., Paik, I. and Ryu, K. H. (2015), Short-term electricity load and price forecasting based on clustering and next symbol prediction. IEEJ Trans Elec Electron Eng, 10: 175–180. doi:10.1002/tee.22050
6. Wen Shen, Vahan Babushkin, Zeyar Aung, and Wei Lee Woon. 2013. An ensemble model for day-ahead electricity demand time series forecasting. In Proceedings of the fourth international conference on Future energy systems (e-Energy '13). ACM, New York, NY, USA, 51-62.
7. Florian Ziel, Rick Steinert, Sven Husmann, Efficient modeling and forecasting of electricity spot prices, Energy Economics, Volume 47, January 2015, Pages 98-111, ISSN 0140-9883
8. Shayeghi, H., Ghasemi, A., Moradzadeh, M. et al. Soft Comput (2017) 21: 525.
9. Neupane, Bijay. Ensemble Learning-based Electricity Price Forecasting for Smart Grid Deployment. Diss. Masdar Institute of Science and Technology, 2013.
10. Liu, Heping, and Jing Shi. "Applying ARMA–GARCH approaches to forecasting short-term electricity prices." Energy Economics 37 (2013): 152-166.

Předběžná náplň práce v anglickém jazyce

Research question and motivation
The amount of renewable energy installed in Europe rose substantially in the last decade. This lead to the ever-growing intraday market to become more important over time. It is still far smaller and less important than the day-ahead market but the rise in its importance is unlikely to stop. Thus the “real-time” forecasting in the electricity market is becoming more and more important as the transactions within the electricity spot-market can take place sometimes even less than an hour before the delivery. The goal of this thesis is to compare an interesting hybrid model to a proven benchmark model.
There has been a substantial amount of literature written in the last decade or so. Ever since the electricity markets got deregulated this kind of research started to matter and have a practical application. Weron (2014) has written a very interesting summary about all of the state-of-art models and methods used not just in short-term but also in medium- and long-term forecasting. Álvarez et al. (2011) proposed in his paper the Pattern Sequence based forecasting and Che & Wang (2010) have proposes in their article the hybrid SVRARIMA model. There are dozens of papers written on this topic and they combine all possible machine learning and econometric models. They differ whether they use more data than just the previous prices (e.g.: Neupane 2013) or not (e.g.: Álvarez 2011, Che & Wang 2010) and whether they use single method or an ensemble learning method (e.g.: Shen et al. 2013) or whether they use some kind of data preprocessing and optimization algorithm (e.g.: Zhang et al. 2012)
Contribution
The main contribution of my work should lie in comparison of SVRARIMA model to PSF model. The hybrid model combining Supporting Vector Machines and Autoregressive Integrated Moving Average model has been used before but was never directly compared to one of the more favorite benchmark models, which is the PSF.
Methodology
The existing SVRARIMA model will be used to forecast the electricity prices from the NordPool area and these results will be compared to the benchmark PSF model using root mean squared error and mean absolute percentage error.
Outline
1. Introduction + motivation
2. Methodology
i. Previous studies
ii. Data
iii. Models
3. Results
i. Comparison of both models
ii. Discussion of results
4. Conclusion