Case Study: JP Elektroprivreda Srbije (EPS) achieves faster, more accurate demand forecasting and higher trading profits with Microsoft Power BI

Serbian energy provider boosts trading profits with predictive AI

JP Elektroprivreda Srbije (EPS), Serbia’s state-owned electric utility, faced slow, error-prone consumption forecasting that increased costs in hourly balancing markets. To modernize forecasting and speed trading decisions, EPS engaged Microsoft Power BI and partners to trial a machine-learning solution built on Azure, Azure Machine Learning, Power Apps and Power BI.

The Azure-based implementation automated forecasts using 20 years of data, with Power Apps streamlining data entry and Microsoft Power BI providing real-time dashboards. Forecasting time dropped from two hours to 15 minutes, error margin fell to 1.7% (from 5–15%), and EPS reports up to EUR 600,000 annual savings on the balancing market plus about EUR 300,000 in additional trading profits.

JP Elektroprivreda Srbije (EPS)

Danilo Komatina

Principal Engineer