Improving Crop Yield with Big Data

April 10, 2018

It’s perhaps the most daunting decision a farmer faces each season: Which of the scores of seed varieties should I plant? Different varietals perform differently in different weather and in different soil conditions at different farms.

The farmer’s investment puts the process in motion irrevocably—like a roll of the dice. And even beyond the borders of the farm, improving crop yield is a critical component for improving food security in challenged populations and protecting natural resources for future generations.

The challenge? Helping farmers, who work with limited resources, harness the power of science and technology to make the best decisions each season to maximize their ROI and improve crop yields.

That’s the focus of Olin Award-winning research from Lingxiu Dong and Durai Sundaramoorthi. Their research outlines an analytics-based framework designed to optimize the farmer’s crop yield each season.


Web app optimizes soybean varietal decisions

“The decision framework … enables farmers to optimally decide the mix of varieties they want to grow based on their risk-tolerance level.”

Taking that research a step further, Dong and Sundaramoorthi developed a web-based application they call “SimSOY,” allowing growers to simulate crop yields for specific soybean varieties under different weather conditions at any target farm.

“The simulation enables the farmer to experiment and assess the risk associated with growing a variety,” the researchers wrote in their paper. “The decision framework … enables farmers to optimally decide the mix of varieties they want to grow based on their risk-tolerance level.”

The researchers did it by evaluating soybean crop data—gathered between 2008 and 2014 from more than 350 sites—involving more than 34,000 data points from 182 varieties. Their analysis compared the performance of different soybean varietals under different weather conditions.

The analysis led to a framework for determining the soybean mix that minimizes risk and maximizes yield at any given farm. Their framework isn’t limited only to soybeans. In fact, it could also help solve problems beyond the realm of agriculture, including nurse-to-patient ratios in hospitals, grocery store product pricing, financial portfolio optimization, and the dynamics of the Standard & Poor’s 500 stock index.

KEY TAKEAWAYS for Managers

  • Using big data from hundreds of soybean crop varieties and planting conditions, researchers developed a simulation to determine the optimal soybean mix.
  • Agribusiness can use this framework as a “decision support tool” to tailor seed-mix recommendations for farmers.
  • The researchers further introduced “SimSOY,” a web-based simulation tool to simulate the yield of a soybean variety under different weather conditions at any target farm.
  • This framework is also applicable to other problems such as nurse-to-patient assignment, pricing of products in grocery stores, portfolio optimization in finance, simulation of ground-level ozone, and predicting dynamics of the S&P 500.
Lingxiu Dong and Durai Sundaramoorthi presented their award-winning research to alumni and friends in the business community on May 8, 2018.

“Machine Learning Based Simulation and Optimization of Soybean Variety Selection”


Lingxiu Dong, PhD, Professor of Operations & Manufacturing Management and Durai Sundaramoorthi, PhD, Senior Lecturer of Management


Under review at Journal of Production and Operations Management