Keywords: Precision Treatment, Average Treatment Effect, Matching
Precision agriculture technology aims to increase crop yields by providing site-specific management tools. At The Climate Corporation, we are developing digital agriculture algorithms to prescribe different management insights based on variability within a field. To evaluate the treatment’s effectiveness, we conducted a trial with more than100 fields by comparing the yields from the Climate FieldView research algorithm with the Grower Choice treatment algorithm. The strip trial was used as an experiment unit for comparison. We applied matching techniques, paired t-test and causal inference to analyze this data. The true results were hidden by adding an artificial signal and a random noise to the original data in order to preserve anonymity. This talk will focus on the lessons learned through this research and analysis. We will discuss the hypothesis formulation, the experimental design, and the statistical methods of evaluating a digital algorithm for precision agriculture technology. This abstract is intended to be presented in a joint session with two other talks from The Climate Corporation.