This paper presents the implementation and analysis of a machine learning-based framework designed to predict crop yield and recommend integrated fertilizer applications using extensive soil, water, crop, and weather data. Leveraging historical datasets, the study applies data pre-processing steps and clustering techniques to identify meaningful patterns relevant to fertilizer and crop recommendation. Various machine learning algorithms including decision trees, random forests, support vector machines, and neural networks are deployed and evaluated. The results demonstrate significant improvements in prediction accuracy and provide actionable insights for optimizing agricultural strategies. This integrated approach supports sustainable farming by enabling precise resource allocation, reducing fertilizer misuse, and enhancing crop productivity, thereby contributing to food security and environmental conservation.