CRC Lecture Series - Machine learning in agricultural economics

Lecturer: Storm, Hugo

This lecture on machine learning in agricultural economics will cover two main aspects.

In the first part, I will discuss what machine learning brings to agricultural economics, 2) what is different in agricultural economics compared to general machine learning, and 3) what are crucial aspects to consider. The discussion is kept at a relatively high level and is intended to be useful for multiple disciplines within DETECT. Agricultural economics should understand under which conditions machine learning tools might be helpful in their work, machine learning experts should understand what is different/specific about our use cases, and other DETECT disciplines should be able to relate the discussed methods/challenges to approaches in their discipline.

In the second part, I will cover approaches to include theory and domain knowledge in data-driven methods. Specifically, I will cover Bayesian Probabilistic Programming as an approach to this. The approach is relevant for our work in DETECT B04. The examples are on estimating yields and crop growth models (building on an example from a crop model presented in the DETECT lecture from Thomas Gaiser). Hence, they should be relevant for various DETECT disciplines.