Published Paper Details:

This research presents an integrated AI framework addressing two critical challenges in Indian agriculture: (1) crop price forecasting for 2026 and (2) data-driven crop recommendation based on soil mineral composition. Using 16 years of historical price data (2010-2025) for 16 major Indian crops, we implement LSTM neural networks and compare with traditional SARIMA models, demonstrating LSTM superiority with 15-25% higher accuracy (R2 = 0.88 vs 0.82 for SARIMA, RMSE reduction of 18.7%, MAPE improvement of 29.3%). The LSTM model generates month-by-month 2026 price predictions with trend analysis (UP/DOWN/FLAT), capturing complex non-linear market dynamics. Simultaneously, we introduce a novel Soil Mineral-Based Crop Recommendation System using Random Forest classification that analyzes soil nutrients (Nitrogen, Phosphorus, Potassium, pH, Organic Carbon) and environmental conditions to recommend optimal crops for specific soil profiles, achieving 96.8% classification accuracy. This dual-pipeline system enables farmers to: (a) identify suitable crops based on soil properties through machine learning, and (b) optimize selling timing through AI-driven price forecasts. Together, these two components form a comprehensive decision-support system enabling farmers to transition from reactive, experience-based agriculture to proactive, data-driven decision-making, ultimately supporting income stabilization and food security objectives across India's agricultural landscape.