# Personalizing Sustainable Agriculture with Causal Machine Learning

**Authors:** Georgios Giannarakis, Vasileios Sitokonstantinou, Roxanne Suzette Lorilla, Charalampos Kontoes  
**arXiv:** [2211.03179](https://arxiv.org/abs/2211.03179)  
**Date:** 2022-11-06  
**Summary page:** https://www.summarizepaper.com/en/paper/2211.03179v1/

## Key points

- Enhancing global crop production is crucial for combating climate change and supporting the growing global population.
- Sustainable intensification in agriculture can help achieve this by transforming agricultural practices from carbon emitters to carbon sinks.
- Understanding the environmental impact of different agricultural management practices is essential for achieving this transformation.
- The global agricultural landscape is highly diverse, requiring personalized agricultural practices based on local conditions.
- Personalized sustainable agriculture can be formulated as an estimation task using Causal Machine Learning techniques.
- Leveraging climate data and land use information, researchers estimated the effects of sustainable practices on Soil Organic Carbon (SOC) content in Lithuania.
- This data-driven perspective provides valuable insights for targeting sustainable practices and expanding the global carbon sink.
- Tailoring sustainable agriculture strategies based on local conditions is important for policymakers and stakeholders promoting environmentally friendly practices.
- Causal machine learning techniques can contribute to achieving personalized sustainability goals in farming.

## Summary

In order to combat climate change and support the growing global population, it is crucial to enhance global crop production. This can be achieved through the concept of "sustainable intensification" in agriculture, which involves transforming agricultural practices from being carbon emitters to becoming carbon sinks. Achieving this transformation requires a comprehensive understanding of the environmental impact of different agricultural management practices and recognizing that the global agricultural landscape is highly diverse. To effectively promote sustainable agriculture and improve green metrics, personalized agricultural practices need to be provided by giving locally adapted management advice. To address this challenge, researchers proposed formulating personalized sustainable agriculture as an estimation task for Conditional Average Treatment Effects using Causal Machine Learning techniques. By leveraging climate data and land use information, as well as employing Double Machine Learning methods, they estimated the heterogeneous effects of sustainable practices on field-level Soil Organic Carbon (SOC) content specifically in Lithuania. This data-driven perspective provides valuable insights for targeting sustainable practices and effectively expanding the global carbon sink. The findings have significant implications for policymakers and stakeholders involved in promoting environmentally friendly agricultural practices worldwide. They highlight the importance of tailoring sustainable agriculture strategies based on local conditions and demonstrate how causal machine learning techniques can contribute to achieving personalized sustainability goals in farming.