From Fog Nets to Neural Nets

Model the water output from water-collecting fog nets in Southwest Morocco. Accurate predictions can improve collection efforts and enable greater access to fresh water throughout the year. #development

$15,000 in prizes
may 2016
543 joined


Dar Si Hmad (DSH) manages a network of fog nets that collect and disseminate fresh water to landlocked communities in Southwest Morocco. Their challenge: how much water can we expect from these nets, and how do we visualize the way these nets are changing lives?

Why

Water in southwest Morocco is scarce and transporting water is difficult. Women and girls spend up to four hours a day to collect poor quality water and carry barrels back to their communities. Dar Si Hmad (DSH) manages fog collection nets that capture and disseminate the fog that rolls over nearby mountains. Managing water dispersal and communicating the system with the community is a difficult task, and essential to providing a dependable solution.

The Solution

DSH had assembled years of data about weather patterns and water yield. In partnership with the Tifawin Institute and Tableau Foundation, DrivenData ran an online tournament to put this data to use. Participants were given two tasks: predict how much water DSH can expect in the future, and create clear and insightful ways to visualize the collections system.

The Results

Top visualizations were judged by a panel of experts and used to foster new conversations with members of the communities that DSH serves. The visualizations that resonated most with each of these audiences were selected for prizes. In addition, the top algorithm improved 25% over the previous benchmark in accurately predicting freshwater output from the fog nets. Measures of leaf wetness and humidity were found to be most useful in making these predictions.


RESULTS ANNOUNCEMENT + MEET THE WINNERS

WINNING MODELS ON GITHUB