The IIF Summer School is (normally) a two-day course which provides in-depth analysis of a cutting edge topic in forecasting from one of the International Symposium on Forecasting (ISF) invited speakers. The Summer School typically shares the same venue as the ISF. Learn more about Virtual ISF 2020–Rio de Janeiro.

Forecasting Summer School 2020 ~ Virtual | Renewable Energy Forecasting – Theory and Practice
November 2-5, 7-10pm GMT
Venue: Virtual
Instructors: Henrik Madsen and Peder Bacher, Technical University of Denmark

The program will be a mix of pre-recorded lecture materials and live lab sessions, where participants can interact with Professors Madsen and Bacher. The program is planned to follow the topics indicated below. However, due to the online nature of the class, the lecturers will focus on methods which have proven to be of importance for forecasting of renewable energy production.

Applications – SUBMIT NOW

There is a class size limit. Therefore, we are requiring applications for attendance. Deadline is September 7, with notifications sent by September 14.

Registration rates (USD)

Non-member: $245
Become IIF member + registration2: $220
Member, local participant (South America), and low income countries: $75
Non-member, student1: $25
Member, student: free

1fee is for a one-year, student membership to the IIF
2fee includes a one-year membership to the IIF ($145 value)
memberships are not refundable

About the Instructors

Henrik Madsen

Henrik received his PhD in Statistics at the Technical University of Denmark in 1986. He was appointed Asst. Prof. in Statistics in 1986, Assoc. Prof. in 1989, and Professor in Mathematical Statistics, with a special focus on Stochastic Dynamical Systems in 1999. In 2017 he was appointed Professor II at NTNU in Trondheim. His main research interest is related to analysis and modelling of stochastic dynamics systems, which includes signal processing, time series analysis, identification, estimation, grey-box modelling, forecasting, optimization and control. Since 1992 he has been the leader of one of the most active research groups in Europe in relation to wind and solar power forecasting as well as methods for integration of renewables in the power systems.

Henrik has received numerous awards, including the Knight of the Order of Dannebrog by Her Majesty the Queen of Denmark, in June 2016. He was appointed Doctor HC at Lund University in June 2017.

In addition, he has authored or co-authored approximately 550 papers and 12 books. The most recent books are Time Series Analysis (2008); General and Generalized Linear Models (2011); Integrating Renewables in Electricity Markets (2013), and Statistics for Finance (2015).

Peder Bacher

Peder is an Associate Professor at the Technical University of Denmark working with statistical modelling of energy systems for forecasting, control and performance assessment. His PhD was in applying statistical models for solar energy applications, which he finished in 2012 and since the aim of his work has been to find good solutions for integrating renewables into the energy system. Finding suitable models for, on the generation side: models for wind, pv and solar thermal forecasting, and on the load and storage side: models and optimization for energy efficiency and demand side management in building HVAC systems, district heating and for battery charging.


November 2-5, 2020 | 7-10pm (GMT) each day

Renewable Energy Forecasting – Theory and Practice

Today, on average, roughly 50 percent of the electricity in Denmark is generated as wind and solar power. Wind power alone accounts for around 44 percent of the electricity load, but this is highly fluctuating. Denmark has hours with almost no wind, but also experiences periods with up to 140 percent of the electricity load. Therefore, forecasting is crucial in order to operate the energy systems including the electricity grid. Prof. Madsen and his collaborators are responsible for the methods used, eg. in Denmark, by both transmission system operators and low voltage operators.

Through a combination of lectures and lab sessions, this course will provides an introduction to methods and tools used for forecasting wind and solar power generation. We will touch upon how the forecasts are used in the daily operation of the power system. For the lab session, R software packages will be used. The topics covered are:

  • Point forecasts and probabilistic forecasts
  • Use of meteorological (MET) forecasts
  • Simple parametric models for forecasting (Box-Jenkins, SARIMA, Holt-Winter, Neural Networks, AI, Hidden Markov, Regime based models)
  • Non- and Semi-parametric methods (Kernel, Spline, Local polynomial, and Varying coefficient based methods)
  • State spaces models in discrete and continuous time
  • Multivariate probabilistic forecasting
  • Methods for forecast evaluation
  • Spatio-temporal forecasting
  • Forecasting hierarchies
  • Combined forecasting (eg. for use of several MET providers)
  • Down- and upscaling
  • Adaptive forecasting
  • Generating forecasts for optimal decision making
  • Tools for wind and solar power forecasting

The course will also provide options of the best tool depending on the penetration level of the renewables and the setting in general.