Interpretable machine learning¶
This section provides the workflow to conduct yield gap decomposition using interpretable machine learning techniques. Machine learning lies at the intersection of computer science and statistics and has been used to unravel patterns in large datasets that are challenging to analyze with more conventional statistical approaches. In agriculture, machine learning is being increasingly used to predict crop yield, although the research community remains divided on which methods are most appropriate given different data types and contexts. With this workflow, you will learn how to (a) perform comparisons of machine learning models, (b) conduct recursive feature elimination, (c) tune models hyper-parameters, (d) assess model performance, and (e) fit and interpret a tuned machine learning model. Estimation of Shapely values will be explained at the end. A companion slide deck can be assessed here.