Microgreen Yield Prediction Under Different Lighting Conditions

S. Desikan, Thaswin Raj Pillai (2024) Microgreen Yield Prediction Under Different Lighting Conditions. Final Year Project (Bachelor), Tunku Abdul Rahman University of Management and Technology.

Text Thaswin Raj Pillai_Full Text.pdf Restricted to Registered users only Download (3MB)

Abstract

The cultivation of microgreens, small leafy plants harvested shortly after the cotyledon leaves develop, has surged in popularity due to their high nutrient density and rapid growth cycle. Given the expanding urban agriculture landscape, there is a critical need to optimize microgreen production to meet increasing consumer demand effectively. This project addresses the technical limitation of predicting microgreen yield under varying light conditions, a significant barrier in scaling up urban agriculture operations efficiently. The primary objective of this research is to develop a predictive model using machine learning algorithms that can accurately forecast microgreen yields based on different lighting conditions. The scope of this study focuses on three popular microgreen varieties such as mung bean, red amaranth, and broccoli cultivated under controlled light spectra, intensity, and photoperiods. The methodology employs a robust experimental setup where microgreens are subjected to varied lighting conditions, and growth data are meticulously collected. This data serves as the foundation for training three machine learning models: Random Forest, K-Nearest Neighbors, and Artificial Neural Networks. These models are evaluated based on their accuracy in predicting yields, using metrics such as Mean Squared Error (MSE) and R-squared values. Significant results from this study indicate that the machine learning models, particularly the Random Forest, show high predictive accuracy, suggesting that precise manipulation of light can significantly influence microgreen yield. The findings demonstrate that specific light conditions optimize growth and yield for different microgreen varieties, with the potential to enhance both the economic and nutritional value of the crops. In conclusion, this research successfully bridges the gap between precision agriculture and urban farming, providing a scalable approach to microgreen cultivation. By integrating machine learning techniques, this project contributes significantly to the optimization of urban agriculture practices, offering a model for future research and practical application in sustainable agriculture.