Machine Learning in Agriculture
Tim Hammerich
News Reporter
Farming is a business, and keeping the business afloat takes data-driven decision making. Mineral CEO Elliot Grant says machine learning models will usher in a new wave of farm data and insights.
Grant… “I like to say that the best time to start collecting data was 10 years ago and the second best time is tomorrow. And it’s a little flippant but it really is important in agriculture because, we won’t see another year like 2022 or 2021 or 2020, and so the importance of training ML models is the diversity of the data that you have. And so in any given year, you are limited to the diversity of data you can collect.”
Grant said that because the model picks up on subtle differences in dynamic environments like farms, it must be constantly retrained and recalibrated to reach peak effectiveness.
Grant… “This never ends, right. ML, it’s not a once and done tool; the technical term is they drift, meaning that if I train a model this year, next year, it’s performance will be slightly worse, the year after will be worse again. Because of these subtle changes in the underlying data, the model will need to be continually retrained and retuned. This is a never-ending process of continuous learning, which means that the technology needs to be continually collecting data, verifying its accuracy, retraining the models, and then deploying the models out into the field.”
While ML is still in its beginning stages of implementation in agriculture, there are exciting hopes for the potential it will have going forward.