AI-Powered Adaptive Fertilizer Recommendation System Using Soil And Weather Data

Abstract

Advances in AI-powered precision agriculture have enabled adaptive fertilizer recommendation systems that integrate real-time soil and weather data to optimize crop nutrition. This paper presents a comprehensive framework that ingests soil nutrient measurements (e.g. N–P–K levels, pH, moisture) and weather forecasts (temperature, precipitation, humidity) to drive machine learning models for site-specific fertilizer guidance. The proposed system leverages publicly available datasets and sensor networks, with algorithms such as gradient-boosted trees achieving up to 99% accuracy in recommending appropriate fertilizer application rates. In simulated evaluations and literature-based experiments, this approach reduced fertilizer usage by ~10% while maintaining yield, demonstrating significant environmental and economic benefits. Key contributions include integrating soil–weather inputs, using explainable ML for model interpretability, and validating performance on real-world data.