Causal forest approach for site-specific input management via on-farm precision experimentation

Estimating site-specific crop yield response to changes input (e.g., seed, fertilizer) management is a critical step in making economically optimal recommendations. Past studies have attempted estimate functions using various Machine Learning (ML) methods, including the Random Forest (RF), Boosted (BRF), and Convolutional Neural Network (CNN) methods. This study proposes use of Causal (CF) model, which one emerging ML methods that comprise “Causal Learning.” Unlike previous yield-prediction-oriented CF focuses strictly on estimating heterogeneous treatment effects (changes yields result from application rates) inputs. We report results Monte Carlo simulations assuming production scenarios test effectiveness nitrogen rates (EONRs), comparing with RF, BRF, CNN. CF’s estimations EONRs were superior under all considered. also show model’s prediction accuracy need not imply EONR accuracy.