Do you have historical yield prediction data ranging back so far that there are enough data points to include in another machine learning model?
I've heard that CropProphet (https://www.cropprophet.com) has 30+ years of forecasts created using statistical methods.
Thank you for your question. We have daily in-season forecast yield for US corn from the 2016 season onward, Argentine soybeans from the 2017 season onward, and US soybeans from 2018 onward.
Our model is trained on the yield data and other variables from 2000 onward, but we do have the actual annual yield data of various sources per crop country dating back as far as the early 20th century.