Skip to main content
Knowledge4Policy
Knowledge for policy
Supporting policy with scientific evidence

We mobilise people and resources to create, curate, make sense of and use knowledge to inform policymaking across Europe.

  • Publication | 2021
Yield performance estimation of corn hybrids using machine learning algorithms
Global Food and Nutrition Security

Highlights:

  • Predicting performance of hybrids prior to field evaluation is a challenging problem.

  • We developed machine learning models to estimate performance of untested hybrids.

  • The XGBoost model achieved the best performance among our proposed models.

  • High-yielding hybrids can be identified by employing our approach.

Abstract:

Estimation of yield performance for crop products is a topic of interest in agriculture. In breeding programs, we cannot test all possible hybrids created by crossing two parents (inbred and tester) since it would be too time consuming and costly. In this paper, we exploit different machine learning algorithms including decision tree, gradient boosting machine, random forest, adaptive boosting, XGBoost and neural network to predict the yield of corn hybrids using data provided in the 2020 Syngenta Crop Challenge. The participants were asked to predict the yield of missing hybrids which were not tested before. Our results show that the prediction obtained by XGBoost is more accurate than other models with a root mean square error equal to 0.0524. Therefore, we use XGBoost model to estimate the yield performance for untested combinations of inbreds and testers. Using this approach, we identify hybrids with high predicted yield that can be bred to increase corn production.