A Multi-Modular AI Approach to Sustainable Agriculture: Integrating Predictive Analytics for Crop Selection and Nutrient Optimization

Abstract

Agriculture remains the backbone of rural livelihoods, with 60% to 70% of the population depending on it. However, farmers often face challenges in making informed decisions due to time constraints, lack of expertise, and limited access to real-time information. This paper explores a comprehensive, web-based decision support system which enables farmers to make real-time predictions and recommendations dependent on the environmental and soil factors such as Nitrogen (N), Phosphorus (P), Potassium (K), pH, temperature, humidity, soil moisture, and rainfall. The application is going to be an Ensemble Model, based on Majority Voting Mechanism, to be applied in crop recommendation and also a Fertilizer Dictionary to provide better fertilizer recommendations. The system based on the Kaggle dataset employs a Sequential Convolutional Neural Network (CNN) for identification of pests. It analyzes market trends for recent past history, historical prices, and provides insights into keeping crops healthy through monitoring in real-time. The application also has an inventory management system that ensures the availability of resources. From crop selection to optimal fertilizer usage to pest control, it promises farmers data-driven insights into increasing productivity in their agricultural outputs with minimal operational risks, offering support for sustainable farming as well.