Thanks -- looking for any and all input. At the Big Data conference this summer a bright farmer talked about his PF work. Among other things, he showed his management zones for one field. They were hand-drawn. I am thinking there must be a way to handle this task algorithmically...just trying to get some parameters. Grouping by quantiles (5, 7, etc.) or looking for natural breaks would delineate areas by yield. But you might have a 10-acre patch in one part of your field and a quarter-acre piece in a corner. You might want both a maximum and minimum size for a sample area? Within reason, I'd think you would want compact zones rather than long, skinny ones. |