Predicting Green Bean Harvest and Yield

Using predictive models to predict green bean harvest, yield and quality to better align supply and demand.
Project complete

In Partnership With:

Mulgowie Fresh Pty Ltd
Queensland University of Technology (QUT)
QLD Government

Predicting Green Bean Harvest and Yield

The Challenge:

Suppliers of fresh produce need to make fast, significant, and complex decisions. There are many complexities in supply and production planning in fresh produce systems with many variables to consider for accurate forecasting of yield, including changing weather patterns.

Planning the supply of fresh produce to align with demand and market expectations is complicated, and with variable data can be costly in term of profitability. Variations in forecast yield can result in supply shortages or excess.

Australian green bean growers need to know when beans can be harvested and what is the likely yield. Managing green bean supply continuity is also important for enhancing brand reputation, value, and growth potential.

An accurate predictive growth model that can predict harvest timing, yield and quality will enhance production and sales throughout the entire business, ensuring customer satisfaction.

The Solution:

This research project used data science and real-world testing to create accurate predictive models that will help the green bean industry to meet market demand.

The project team created a dedicated green bean modelling tool for Mulgowie Farming Company by identifying and building on existing crop planting, yield, and location data.

The tool was then commercially tested and further refined throughout the 12 month hyper-care period.

The predictive model will help Mulgowie Farming Company and the green bean industry to make better production and sales decisions based on harvest timing, yield, and quality, while supporting marketing, promotion and pricing.


The collaborative nature and commercially focused and implemented outputs from this research demonstrates the potential for problem solving across other agricultural sectors and within individual business and business group

This initiative assists and enhances all aspects of the Mulgowie farming Companies production operations across all their Australian production and packing sites.

Key results:

  1. Analysis of predictive model performance shows competitive level of accuracy with current practices, with avenues for further improvement over time. Current practice for pre-harvest human crop yield assessments will continue to be important and will further enhance predictive model accuracy and development.
  2. The production planning algorithm demonstrates good capability to match production to sales targets within the constraints of the current Mulgowie system. The research team will continue to support Mulgowie in adoption and refinement of the technologies developed in this project.
  3. The infrastructure, methods and processes developed in this project provide a valuable platform and example for ongoing improvement and development of underlying models and algorithms in the crop growing and broader agricultural sector.

Read the final project report.

meet the team

Majella Nolan

Innovation Manager, Food Agility CRC

David Carey

Senior Horticulturist, Department of Agriculture and Fisheries Queensland

Associate Professor Paul Corry

Associate Professor in Operations Research, Science and Engineering Faculty, School of Mathematical Sciences, Queensland University of Technology

Amanda Woods

Project Lead, Mulgowie Farming Company


No items found.

other projects

related projects

No items found.