Machine Learning Solution for Urban Agriculture Planning

Environmental Science • Milan Project

Challenge

Milan faces increasing pressures to improve urban sustainability, food security, and environmental resilience. This project tackled multiple challenges, including identifying suitable areas for urban agriculture, predicting crop yields, and managing risks related to pests and plant diseases. The goal was to provide data-driven insights to maximize agricultural productivity and sustainability within the city.

Approach

As the lead for the exploratory data analysis (EDA) on yield prediction, I spearheaded the effort to process and analyze key environmental variables such as NDVI (vegetation index), soil quality, and weather data. Using machine learning, we developed models to predict crop yields in different regions. In parallel, we applied clustering algorithms to identify the most suitable areas for urban agriculture by analyzing environmental and geographical factors. Additionally, we evaluated potential risks of pests and plant diseases to ensure comprehensive planning for sustainable urban agriculture.

Results

The project identified optimal zones for urban agriculture in Milan and provided accurate crop yield predictions based on environmental conditions. The clustering analysis allowed us to pinpoint high-potential areas for farming, while the pest and disease management models highlighted risk zones for early intervention. These findings were presented in a clear, actionable dashboard for city planners to inform strategic decision-making around urban agriculture, crop management, and sustainability efforts.

Future Plans

The next phase involves enhancing the model by integrating more detailed real-time data on pests and diseases and extending the analysis to other urban areas. Future work will focus on improving long-term yield predictions by incorporating climate forecasts, further increasing the resilience and efficiency of urban agriculture initiatives.

Expertise

As the lead on data analysis for yield prediction in an urban agriculture project, I guided the team’s approach to understanding crop productivity through advanced statistical analysis and yield modeling techniques. In addition to driving the yield analysis, I applied clustering, geospatial analysis and feature engineering to identify optimal areas for urban farming, leveraging multiple data layers to support crop planning. This combination of data-driven insights allowed us to address the complex needs of urban agriculture effectively.