Energy Fingerprints: What Can You Learn from Electricity Data?

The energy transition is driving the deployment of electric vehicles, heat pumps, rooftop solar systems, batteries, and other distributed energy resources. However, companies like AEW often only have access to aggregated smart-meter measurements and lack visibility into what happens behind the meter. 

In this challenge, participants will analyze multiple years of anonymized 15 minute electricity consumption data to identify energy assets and usage patterns within individual households. Can you determine whether a customer owns an EV, operates a heat pump, produces solar energy, or uses other characteristic devices? Can you estimate when these assets are active and how their behavior changes over time? 

The objective is to develop machine learning and pattern recognition approaches that extract meaningful insights from load profiles while preserving customer privacy. Participants are encouraged to combine signal processing, feature engineering, time-series analytics, and modern AI techniques to uncover the hidden "energy fingerprints" of households. 

The resulting models should provide interpretable insights and scale to large customer populations. Accuracy, robustness, explainability, and innovation will be key evaluation criteria.

We provide a multiple year timeseries storing.

• Data Science
• Machine Learning