Research Question: Using measurements of environmental variables such as rainfall, temperature and humidity for a specific set of agricultural regions across the world, can we computationally identify the causal effect of environmental variables to predict agricultural yield and optimize the yield by prioritizing proactive assistance such as increased irrigation or fertilizers?
In this research project, I led a team of student researchers under the guidance of Stanford faculty members to analyze the effect of environmental variables on agricultural yield and to identify proactive actions using machine learning algorithms for optimizing the yield. We also explored various data processing and data analysis methods to gain insights from various data sources.
One major challenge was the lack of high-quality, consistent data across various environmental factors, such as rainfall, temperature, and soil health. To address this, we merged various data sets used data augmentation techniques to fill gaps. Additionally, designing a causal inference model that accurately captured the relationship between the variables and yield was complex. We addressed this by leveraging advanced statistical methods and iterative testing to refine our models.
The machine learning models successfully identified key environmental factors influencing agricultural yield. The optimization model provided actionable recommendations for farmers, enabling them to improve crop yield by making data-driven decisions. The results were presented in a comprehensive report and shared with faculty and student researchers, demonstrating the potential for AI to enhance agricultural practices.