SAS validates whale protection program with machine learning

Data and AI leader SAS is helping Fathom Science Inc., a North Carolina State University tech spin-off that is building digital twins of the ocean, validate their state-of-the-science whale location prediction model so vessels can avoid striking critically endangered North Atlantic right whales. Fathom Science seeks to provide expert, tailored information on the marine environment to enable timely and reliable decision making, empowering transformative marine operations and conservation efforts.

To help protect the whales, Taylor Shropshire, Head of Marine Resiliency at Fathom Science, combined analysis of historical whale sightings with Fathom’s ocean model to create WhaleCast, a heatmap projection showing the likelihood of right whale activity along the East Coast. It is designed to be integrated with vessels’ existing on-board touch screens so mariners can gauge the likelihood of right whales’ presence. Ships can use this information to reduce speed in flagged areas with the goal of decreasing the strike rate on these endangered whales.

Shropshire sought validation of WhaleCast through deeper statistical and machine learning methods, so he turned to SAS’ Data for Good program.    

For the SAS Data for Good volunteer to explore machine learning models using SAS^® Viya^®, he needed additional data. Using SAS^® Data Maker, he generated synthetic data with characteristics closely matching the original data for more effective validation. Now armed with nearly 500,000 data points, the volunteer was able to divide the data into groups to perform more robust training, validation and testing for seven machine learning models.

Following the initial modeling of WhaleCast, the volunteer used SAS Viya Workbench, a standalone programming environment, to help solve a second problem: calculating the probability of whales’ distance from shore. In Viya Workbench, the volunteer was able to quickly program calculations and add the distance data to the models he had created, further validating the accuracy Fathom’s approach.

“It was interesting to see SAS create multiple models very quickly,” Shropshire said. “They were able to expand from a very simple model to complex neural network and machine learning-type models to show the benefits and the limitations of each.”

Learn more on how Fathom Science is using SAS to protect marine life.  

SAS is a global leader in data and AI. With SAS software and industry-specific solutions, organizations transform data into trusted decisions. SAS gives you THE POWER TO KNOW^®.