Power Laws: Detecting Anomalies in Usage

Commercial buildings waste an estimated 15% to 30% of energy used due to poorly maintained, degraded, and improperly controlled equipment. Competitors built quick-response algorithms to find anomalies in energy use and elevate them for human attention and intervention. #energy

€26,000 in prizes
Completed mar 2018
649 joined

alt-text

Each challenge in the Power Laws series explored a different aspect of energy efficiency and management. The winning algorithms were released under an open source license to spread understanding about how energy modeling works and what approaches are most effective.

Why

As energy consumption of buildings has steadily increased, more and more building do not perform as intended by their designers. Typical buildings consume around 20% more energy than necessary due to faulty construction, malfunctioning equipment, incorrectly configured control systems, and inappropriate operating procedures.

The Solution

Automatic, quick-responding, accurate, and reliable fault detection can ensure better operations and save energy. In this competition, data scientists all over the world built algorithms to identify anomalous energy consumption.

The Results

Identifying anomalies was a tricky task, and the best performers combined human judgment with machine suggestions. The winning model was very precise in its predictions (which was highly weighted in this challenge), correctly identifying anomalies in 100% of the observations it predicted to be anomalous.

Ultimately, these ideas can help drive forward the kinds of anomalies that we hope to identify in building energy use and to respond more quickly to energy-wasting defects.


RESULTS ANNOUNCEMENT + MEET THE WINNERS

WINNING MODELS ON GITHUB

DATASET ON SCHNEIDER ELECTRIC EXCHANGE