That’s the view of Australian technologist Chris Mendes who says growers need a data model.
“You’ve got to be able to actually say ‘this is what my data looks like to the computer. These are the things that matter. This is how they relate to each other’,” he explained.
“But if you want to share amongst yourselves, this is going to be essential: you’ve got to have an agreed data model where you can actually exchange information in a meaningful way. You need a platform where you can bring it all into one place.”
Mendes, a founder and chief technology officer of the Australian company Yield, spoke at the Apples and Pears section of the inaugural joint Horticulture Week in Christchurch.
He told the conference that better use of resources on the ground, the better and earlier predictions of harvest timing and yields will ultimately interact to increase productivity.
Mendes, an electrical engineer by training, told the conference he had moved into largescale data crunching in the finance sector before being invited to consider how data management could be applied to agriculture.
His company has now produced a yield management platform that answers growers’ questions like how much they’re going to produce, when will the harvest be and when is labour needed, fruit quality, tonnage and other factors.
“All of this feeds into your supply chain and other logistics that you’re going to have to deal with,” Mendes told growers. “It is really about predictive models, it’s forward looking. It’s about recommendations, and what you need to do to be ready for tomorrow.”
Mendes explained when he worked in finance, they were importing a lot of data into a central model but had to build individual translators to handle the data coming from each stock exchange. Yield had now done something similar with the crop management system because growers want to do things their own way.
“It’s going to be a lot easier though if you guys work together and come up with your own model.”
Mendes says growers have a lot of data “floating around” but a lot of it was siloed information.
Prior to launching in the US, the company had surveyed businesses there and found that they overwhelmingly knew that managing their data was their future but they were struggling with it.
A huge number also said that they would like to have all their data in one place.
“This is a really important thing because so many of the businesses that you guys run have got data in all sorts of systems,” he added.
“You’ll have an HR system, you’ll have a harvest management system, you might have a farm management system, all of those things, and the data is kind of isolated and it makes it really difficult for you to pull reports together to understand what is going on in your business.”
Mendes says 75% of the surveyed growers were spending 11 hours or more a week wrangling data, ‘just fiddling around bringing their data together’, which is a huge productivity loss.
He says while machine learning sounds scary, it was about taking the data that growers generate themselves, and applying some methods to it to see what works. Once it’s trained it can be put into production; when it stops working it can be retrained.
“If you’ve got the right data, you can really get quite good predictive models in place for yourself. That’ll give you a lot of early warning about what’s going to happen.”
One example he gave was how the platform assessed the last four years’ weather in the Barossa Valley and was able to predict a late grape harvest for the 2023 vintage.
Mendes says this information was a huge value to the growers.
“We know growers who didn’t use it and they had people waiting on the property to harvest grapes a month before it was ready.
“And, as you can imagine, that cost these growers a lot of money.”
