Integrating Machine Learning and IoT Sensors for Enhanced Soil Nutrient Monitoring and Crop Recommendation Systems

Abstract

With the increase in the number of IoT farming datasets, identifying the appropriate data for IoT agriculture applications has become increasingly challenging. This research presents an advanced crop recommendation system developed by integrating various datasets, including Crop_Recommendation.csv, Soil.csv, and Crop_names.csv, which provide the foundation for accurate crop predictions. The system leverages geographic coordinates (latitude ϕ and longitude λ) to model environmental factors like temperature and humidity using regression models, forming essential inputs for crop suitability analysis. By applying a classification model , trained on features such as soil type and nitrogen requirements, the system predicts the most suitable crop class . Hyperparameter tuning optimizes the model to ensure robust predictions, and the system ranks the top five crops based on their likelihood of thriving under given conditions. Additionally, the system calculates Growth Degree Days (GDD) and nutrient requirements (nitrogen, phosphorus, potassium) for each recommended crop, offering a comprehensive decision-making tool for farmers. This framework, grounded in machine learning and geographical data, enhances agricultural decision-making by providing precise, data-driven crop recommendations tailored to specific environmental and soil conditions.