Automated Dynamic Machine Learning for time based forecasts (for Energy, Utility & Commodity sectors)

• Research unit: INNOSUISSE
• Project number: 18571.1 PFES-ES

Short description (German)

Automated Dynamic Machine Learning for time based forecasts (for Energy, Utility & Commodity sectors)

Short description (English)

Automated Dynamic Machine Learning for time based forecasts (for Energy, Utility & Commodity sectors)

Abstract (German)

The project develops new algorithms to dynamically automate feature generation process of a machine learning platform during the learning phase, applied to time series. The solution will be used in Energy & Commodity sectors. The key benefits and objectives of this research will be (a) an increase in the speed of calculation by 90 %, (b) an operational cost reduction and (c) an improvement in the prediction accuracy by 20-40% compared to today's state of the art.

Abstract (English)