Mapping wheat crop phenology and the yield using machine learning (ML)

Abstract

Wheat has been a prime source of food for the mankind for centuries. The final wheat grain yield is the multitude of the complex interaction among the various yield attributes such as kernel per plant, Spike per plant, NSpt/s, Spike Dry Weight (SDW), etc. Different approaches have been followed to understand the non-linear relationship between the attributes and the yield to manage the crop better in the context of precision agriculture. In this study, Principle Component analysis (PCA) and Stepwise regression used to reduce the dimension of the original data to get the critical attributes under study. The reduced dataset is then modeled using the Radial Basis neural network. RBNN provides the regression value more than 0.95 which indicates the strong dependence of the yield on the critical traits.