New data-based products to make sheep management easier

Managing the well-being of sheep is set to become significantly easier for producers, thanks to a new collaboration tapping into the data science used in Australia’s national security systems.

Screen Shot 2016-06-01 at 7.23.28 AM

Sheep CRC CEO James Rowe; D2D CRC CEO Sanjay Mazumdar; D2D CRC Research Director, Brenton Cooper; D2D CRC Senior Data Scientist, Troy Wuttke and D2D CRC Machine Learning Specialist, Jason Signolet.

An agreement between the Cooperative Research Centre for Sheep Industry Innovation (Sheep CRC) and the Data to Decisions Cooperative Research Centre (D2D CRC) will see the two groups working together to develop new data-based products to make sheep management easier and achieve better outcomes in terms of sheep well-being and productivity.

D2D CRC chief executive Sanjay Mazumdar said it’s the first time the D2D CRC has collaborated with the agriculture sector.

“The D2D CRC’s main focus has been on defence and national security, however we can take our research and data skills and apply them to other sectors of the Australian economy. Our work is very applicable to any data-intensive industry,” Dr Mazumdar said.

The two CRCs have been working together for the past 18 months under a memorandum of understanding to define opportunities for big data applications in livestock industries and weigh up the benefits of a partnership for both CRCs.

Sheep CRC chief executive James Rowe said the combination of a thorough scoping process and some small pilot projects provided convincing evidence that synergies would arise from collaboration.

“The first major project under this collaboration is the development of a data platform and web-based app to support better management of sheep well-being, due to be completed by early 2017,” Professor Rowe said.

“The app will allow sheep producers to anticipate risks that might compromise the wellbeing of their sheep flocks using climatic data produced by the Bureau of Meteorology in combination with a number of bio-physical models and a range of data science technologies based around machine learning.”

This first major collaborative project comprises three components, with the first module using climate data, satellite images and bio-physical models to predict feed availability as well as risks from parasites, flies and extreme weather events.

“The second aspect deals with livestock data such as liveweight, condition score and pregnancy scanning.  It also includes information about the genetic background of a flock,” Prof. Rowe said.

“The third component is the assessment of risk as a combination of livestock resilience and environmental factors.

“The really novel feature of the data platform for the livestock sector is the ability to automatically update predictions of environmental risks and animal resilience/susceptibility in real-time, without needing labour-intensive inputs from sheep producers.

“The platform utilises an automatic feed of climate data from the Bureau of Meteorology and updates models of soil moisture, feed budgets, parasite risks and extreme weather on a daily basis for individual properties.

“This novel approach is only possible through a number of new developments in the rapidly changing data technology landscape – faster computing, cheaper data storage, cloud-based data technologies and increasingly powerful machine-learning algorithms.

“Getting the most out of these new technologies requires highly specialised skills and an ambitious project of this nature would be impossible without the high-level data science experience of the D2D CRC.

“Just as important as the technical aspects of the project is the challenge of ensuring that the new information systems meet the needs and expectations of our end-users – commercial sheep producers.