FERTICAST – Data-Driven Fertilizer Optimization System Embedded with Rainfall Prediction

Abstract

Fertilizer management is a major challenge in modern agriculture, especially due to changing weather conditions that directly affect soil nutrients and crop growth. Traditional fertilizer recommendations are usually fixed and do not consider factors like rainfall, which can wash away essential nutrients and reduce their effectiveness. This often leads to unnecessary fertilizer usage, increased costs for farmers, and environmental issues such as soil and water pollution. To address this problem, this paper introduces Ferticast, a smart and data-driven fertilizer recommendation system. The system uses a Random Forest Regression model to predict the required levels of nitrogen (N), phosphorus (P), and potassium (K) based on crop type and environmental conditions like temperature, humidity, and rainfall. In addition, a rainfall-aware module analyzes short-term weather forecasts to decide whether it is safe to apply fertilizers or if it should be delayed to avoid nutrient loss. The results show that the system provides reliable and adaptive recommendations, improving fertilizer efficiency compared to traditional methods. By combining machine learning with weather intelligence, Ferticast helps farmers make better decisions. Overall, the proposed system supports sustainable agriculture by reducing waste, improving crop productivity, and promoting smarter fertilizer usage.