|
| Some Saturday afternoon thoughts on number-crunching/PA data
I want to compare the two areas of seed selection and soil sampling/fertility applications.
In both cases, there is a simple method, plant a test plot with several numbers from dealers you are comfortable buying seed from. Take the top five (or 4,6,whatever) numbers and plant them the following year. Repeat the cycle. This is fairly easy, makes good sense, but has lots of shortcomings and I think we can do far better.
Likewise in soil sampling, the easiest method is to go to a field, pull a nice composite sample across the field and fertilize accordingly.
On the seed side, I believe that we can never have too many plots and comparisions to look at, I think we have to understand that the number that performs one year may not the next, we need to consider things like harvest schedule, standability, early harvest premium potential. Above all, we must realize the best approach is to review a lot of data that does not originate from our own operation… In the end, crunching some numbers and getting a broader perspective will serve us well
Likewise in soil sampling, we can more intensely sample the field at some level, we can review samples and results from previous years, we can consider how long we might be farming the ground, we can think about spending our dollars on the farms/areas with the greatest potential for return. Same as seed selection, this all requires crunching a bunch of data. And this may open up a can of worms, but just doing a grid sample and applying a standard rec to it like is so common nowdays, really is not much different than the single composite. I am talking about attempting to use multiple years/layers of data when developing the rec..
It seems to me that the challenge has been that once you start really looking at the data, it gets very, very messy very quickly. So that what was a nice little test plot or soil sample becomes a mess of numbers that don’t make sense. I think that too often, we have gotten to that point, thrown up our hands in disgust, and just said forget it, I’ll just spread on crop removal or a simple average, or I’ll plant whatever seed “feels” right.
Having spent time in both of these areas recently, I am convinced we can make incremental improvements in each, while avoiding the belief/trap that we are going to do anything to “change the world”. Key to making the whole process work is keeping the cost down, which requires proper software/data processing systems in place.
Thoughts?
| |
|
Thinking about data..
