Profiling energy customers for Eesti Energia (2022)
CITIS devised a ML-based methodology and built a custom software solution for the state-owned energy group Eesti Energia that identifies customers’ energy consumption profiles automatically and detects when those profiles change. The core of the work was machine learning: clustering customers by how they actually use energy, rather than by the categories a utility usually has to hand, and flagging when someone’s pattern shifts. We also developed the user interface for reading the results, and did preliminary work on forecasting consumption time series.
The collaboration also produced public policy analysis. In June 2023 Andres Võrk and Helen Poltimäe published a study (in Estonian) in Riigikogu Toimetised, the Estonian parliament’s journal, on how introducing the fixed-price universal service in October 2022 changed household electricity consumption, finding that capping the peak price shifted use back into peak hours.
The project ran from January to December 2022 at the Center of IT Impact Studies, commissioned through procurement, and was carried out by Mihkel Solvak, Peeter Leets, Andres Võrk, and Taavi Unt.
