ADVANCING SUSTAINABLE AGRICULTURAL DEVELOPMENT AND FOOD SECURITY THROUGH MACHINE LEARNING: A COMPARATIVE ANALYSIS OF CROP YIELD PREDICTION MODELS IN INDIAN AGRICULTURE

This paper discusses the crucial issue of improving decision-making in agriculture and ensuring food security, which are key elements of Sustainable Development Goal (SDG) 2: Zero Hunger.Its objective is to eliminate hunger, attain food security, enhance nutrition, and advance sustainable agriculture.Conventional methods for estimating crop yields based on past historical data, weather patterns, and expert insights often lack the scalability and accuracy necessary for sustainable agricultural progress.This investigation delves into the application of machine learning (ML) strategies to forecast crop yields, utilizing a comprehensive dataset from Indian agriculture encompassing factors like fertilizer usage, seasonal fluctuations, and soil compositions.The performance of ten ML algorithms was assessed, including Linear Regression, LASSO, RIDGE, K-Neighbours Regressor, Decision Tree Regressor, Support Vector Regressor, Extreme Gradient Boosting Regressor, AdaBoost Regressor, CatBoost Regressor, and Gradient Boost Regressor.Our results indicate that CatBoost Regressor, K-Neighbours Regressor, and Gradient Boost Regressor demonstrate superior predictive precision, showcasing their potential for practical implementation in the agricultural domain to facilitate the realization of SDG 2. These models present notable reliability and efficacy in predicting crop yields, which are crucial for informed decision-making aimed at optimizing agricultural yields and contributing to sustainable agricultural growth and food security.The study furnishes valuable insights for farmers, policymakers, and supply chain managers, proposing that the adoption of efficient ML models can greatly enhance decision-making processes in agriculture.