Uczenie głębokie w prognozowaniu punktowym, probabilistycznym i trajektorii cen 
energii elektrycznej
[Deep learning in point, probabilistic and ensemble forecasting of electricity prices]

Grzegorz Marcjasz (ORBI, PWr, Wrocław)


This talk will summarize the main results of my PhD thesis, which aims at developing 
robust, reliable and - when possible - interpretable deep learning-based approaches 
for short-term point, probabilistic and ensemble forecasting of electricity prices. 
Approaches that significantly outperform regression-based predictions not only in 
terms of statistical error measures, but also in terms of economic benefits for 
entities involved in intraday and day-ahead trading. I will also highlight the 
importance of following best practices, as robust comparisons and replicability 
are key to research excellence.