Sustainable Industry: Rinse Over Run

Help make industrial cleaning processes more efficient! The goal of this competition is to predict measures of cleanliness during final rinse in order to help minimize the use of water, energy and time, all while ensuring high cleaning standards. #energy

€20,000 in prizes
Completed mar 2019
1,213 joined

alt-text

Each DrivenData challenge with Schneider Electric explores a different aspect of energy efficiency, water conservation, and natural resource management in an era of environmental change. The winning algorithms are released under an open source license to spread understanding of these data challenges and the approaches that are most effective.

Why

Efficient cleaning of production equipment is vital in the Food & Beverage industry. Strict industry cleaning standards apply to the presence of particles, bacteria, allergens, and other potentially dangerous materials. At the same time, the execution of cleaning processes requires substantial resources in the form of time and cleaning supplies.

Better foresight into cleanliness levels at the end of these pipelines can help minimize the use of water, energy and time, all while ensuring high cleaning standards.

The Solution

Schneider Electric ran a machine learning competition to predict levels of turbidity – a standard industry measure of cleanliness – detected in the final stage of cleaning processes. In Stage 1, more than 1,200 data scientists competed over 50 days to build the most reliable predictions. In Stage 2, the 15 best-performing teams were invited to submit brief reports to illuminate business implications of quantitative patterns in the data.

The Results

Final scores were based on mean absolute percent error, indicating how closely predicted turbidity levels matched actual measurements. The winning models achieved scores of just 27% - a dramatic reduction compared to more than 5x the error reported by the pre-challenge benchmark (146%)!

All 15 prize-winning solutions have been shared for continued learning and development.


RESULTS ANNOUNCEMENT + MEET THE WINNERS

WINNING MODELS ON GITHUB

DATASET ON SCHNEIDER ELECTRIC EXCHANGE