Syngenta Crop Challenge plants seeds for success in agriculture

Agriculture is in need of innovation, and industry leaders are increasingly looking to analytics experts to translate real-world crop data into tools to aid decision-making – for both plant breeders and farmers. Arable land and water are being used faster than the planet can sustain, but our population keeps growing. By 2050, there will be an estimated 9.8 billion people globally, so maximizing crop productivity is essential to feed the growing population.

With the Syngenta Crop Challenge in Analytics, sponsored by the Analytics Society of INFORMS, students and professionals in the operations research and analytics community can contribute their insights and skills to help plant scientists breed seeds that produce higher-yielding crops, which can ultimately help improve food security.

“Through our partnership with Syngenta, we provide our student and other members with exposure to a critical real-world challenge – feeding the world’s fast-growing population,” says Erick Wikum, president of the Analytics Society of INFORMS. “The competition not only provides our members with the opportunity to display their analytics chops, but it also motivates them to pursue jobs applying analytics to benefit society.”

Now in its fourth year, the challenge addresses crop production scenarios that can help researchers determine which seeds continue on the path to development and commercialization. Participants receive real-world data to analyze and are free to use the methodology they choose. The insight that can be gained from predicting performance of hybrids based on changing variables such as weather or soil can help plant breeders determine which seed varieties are going to produce the strongest, highest yielding crops.

“We are using big data and analytics to move beyond describing how a product performs, to predicting how a product will perform in various environments,” says Gregory Doonan, Crop Challenge judge and head of novel algorithm advancement, Syngenta. “By improving our predictive models, we can move to a prescriptive mindset and ask the question: What product or set of products should a grower plant on his or her field to maximize performance while also minimizing risk?” 

2019 Crop Challenge

The 2019 Crop Challenge question asked participants to define stress metrics and classify corn hybrids for stress tolerances. Understanding how a hybrid reacts to stress such as drought can help breeders develop seeds that can potentially thrive across a range of environmental scenarios, even in regions that may be less than hospitable for a crop. If farmers are given access to more resilient and localized seed options, they have the potential to grow crops where they couldn’t before.

“I look forward to reviewing this year’s entries,” Doonan adds. “In particular, I want to see how the entrants characterized environments using machine learning and feature engineering. The features that are used are important in the accuracy of predictive models, and I’m excited to review the different approaches.”

The 2019 competition has closed, and entries will be reviewed by a prize committee made up of professionals with extensive backgrounds in analytics and engineering, with focused interests in areas such as machine learning, business analytics, operations research and optimization, supply chain and operations through advanced analytics and artificial intelligence in agricultural research.

Several of the judges have participated in past Crop Challenge contests, including Durai Sundaramoorthi, prize committee vice-chair, who finished third in 2017. His team developed a model that optimized seed selection – research that ultimately resulted in their receipt of the 2018 Olin Award for research that transforms business, from the Olin Business School at Washington University in St. Louis.

Lizhi Wang, a professor from Iowa State University and a Crop Challenge judge, placed third in 2018 when his team designed a method of predicting crop yield. Daniel Jiménez joined the 2019 committee as a judge after leading a team to first place in 2018 with a model that predicted corn hybrid performance using machine learning. 

Beyond the Challenge

In past years, teams have used their Crop Challenge submissions to secure funding to continue their research. In 2016, a team from the BioSense Institute in Serbia placed fourth in the competition and used their entry to secure a large grant from Horizon 2020, a European research group that funds innovative projects. The BioSense team went on to win the 2017 Syngenta Crop Challenge.

In 2018, a team member from The International Center for Tropical Agriculture (CIAT) in Colombia won first place in the Challenge, then used the same research to develop a similar solution to optimize planting to improve farm profitability.

Syngenta and INFORMS have announced plans to continue the Crop Challenge through 2020 so that they can continue to promote collaboration across industries and tackle some of the most daunting challenges in agriculture.

“We look forward to our continued partnership with INFORMS,” says Doonan. “We are not only able to access exceptional talent, we are able to build excitement and foster the next generation of scientists focused on applying analytics to modern agriculture.”