Do you have questions about Data Science?

We’d love to hear from you. 

Contact

Case Study: Forecasting Energy Prices in the Electricity Sector

For a company in the energy sector, COSMO CONSULT developed a model to forecast electricity volumes that takes into account a range of different influencing variables on an hourly basis for a period of one week. Using this method, a forecast quality of over 91% was achieved. This made it possible to manage short-term electricity trading efficiently and to selectively control it.

Key facts

  • Accurate hourly electricity price forecasting
  • Use of machine learning and time series modeling
  • Consideration of seasonal structures
  • Integration of weather data and other external information

Initial situation

For short-term trading in the electricity sector, a company in the energy sector already ran a simple forecasting model that no longer fulfilled the growing requirements for speed and calibration. Consequently, a method of predicting electricity prices more efficiently was sought.

COSMO CONSULT was commissioned to develop a solution that facilitates hourly electricity price forecasting for a period of up to one week using many different influencing variables - with high forecasting quality.

Approach

The model was built using numerous production variables as well as historical weather data and weather forecasts from recent years. COSMO CONSULT was responsible for obtaining this information. Various mathematical algorithms such as ordinary time series models (ARIMA, exponential smoothing), regression models (KRLS or machine learning approaches such as gradient boosting) ensured an optimal result, which was then transferred into a complete model. The process also imported missing values, taking into account seasonal structures and seasonal variations. The project team then used backtesting to validate these values against each other and calibrated the parameters of the entire model. The significance of the weather and other variables was also assessed in order to analyze their influence on forecasting quality.

The result was a solution that automatically forecasts the price of electricity on an hourly basis over a consecutive period of seven days.

Additional benefit

The forecast is provided as a table which can be used in Excel. An overall forecasting quality of over 91% was achieved (MAPE of less than 9%). This allowed the company to manage its short-term electricity trading efficiently and to direct it in a targeted fashion.

Ask our experts

COSMO CONSULT has many years of experience in providing digital solutions in the field of data science. Our services are based on a clear approach, detailed knowledge of business processes, and excellent product expertise. Our experts will be happy to advise you on the unique possibilities available to you when you use modern software technologies. Please give us a call! We look forward to talking with you on how your business can enter the digital age.

Contact Us!

COSMO CONSULT International

COSMO CONSULT SI GmbH
Lothringerstraße 14
1030 Wien
E-Mail
COSMO CONSULT
201 rue Carnot
94120 Fontenay-sous-Bois
+33 1 83010360
E-Mail
COSMO CONSULT
Schoenebergerstr. 15
10963 Berlin
+49 30 343815-0
+49 30 343815-111
E-Mail
COSMO CONSULT SPAIN, S.A.U.
Vallespir, 19 4ª Módulo 2
08173 Sant Cugat del Vallès, Barcelona
+34 902820242
E-Mail
COSMO CONSULT AB
Ralambsvaegen 17
11259 Stockholm
+46 8 7998660
E-Mail
COSMO CONSULT
Ruetistrasse 13
8952 Schlieren
+41 44 8306465
E-Mail
COSMO CONSULT Business S.L.
Presidente Riesco 5711, Oficina 802b
Las Condes, Santiago de Chile
+569 73769126
E-Mail
COSMO CONSULT Business S.L.
Avenida Cra. 45 # 114-78, Piso 6
Bogotá - Colombia
+57 320 7935148
E-Mail
COSMO CONSULT Business S.L
Checoslovaquia E10-105 y Eloy Alfaro
Quito, Ecuador
‪+593 99 608 4180
E-Mail
COSMO CONSULT Business S.L.
Rodolfo Gaona 81 Piso 7
Colonia Lomas de Sotelo
México, D.F. 11200
+52 55 2452-9200
E-Mail
Edificio Magel, Segundo Piso
Ave Samuel Lewis, Urb. Obarrio,
Ciudad de Panamá, República de Panamá
+507 62800343
E-Mail