Smart Agriculture Assistant: An Integrated AI-Driven Decision Support System for Crop Recommendation, Yield Prediction, Fertilizer Optimization and Market Price Forecasting
论文提出集成AI决策支持系统，方法较新且性能指标高，但来源为普通期刊，时效性差，影响范围有限。
印度 Nagarjuna College of Engineering and Technology 的 Abeed K Attar 等提出 Smart Agriculture Assistant 综合 AI 驱动决策支持系统，集作物推荐、产量预测、肥料优化和市场价格预测于一体。系统利用 Random Forest、XGBoost、CatBoost、LSTM 和 SHAP 等多种机器学习算法分析 Kaggle 和 Agmarknet 收集的农业和市场数据集。框架处理土壤养分（N、P、K）、pH、温度、湿度、降雨量、作物特征和历史市场价格数据，为农民生成智能推荐和预测洞察。实验评估表明，所提出的混合框架达到 99% 的准确率，优于传统机器学习模型。
Authors
- Abeed K Attar Department of Computer Science Nagarjuna College of Engineering and Technology Bengaluru Rural, Karnataka, India Author
- Chetan K V Department of Computer Science Nagarjuna College of Engineering and Technology Bengaluru Rural, Karnataka, India Author
- Karthik S Tevari Department of Computer Science Nagarjuna College of Engineering and Technology Bengaluru Rural, Karnataka, India Author
- Sanjay Yogi Department of Computer Science Nagarjuna College of Engineering and Technology Bengaluru Rural, Karnataka, India Author
- Rashmi P Karchi Department of Computer Science Nagarjuna College of Engineering and Technology Bengaluru Rural, Karnataka, India Author
DOI:
https://doi.org/10.47392/IRJAEM.2026.0357
Keywords:
Artificial Intelligence, Crop Recommendation, Ma-chine Learning, Market Price, Forecasting, Smart Agriculture.
Abstract
Agriculture plays a vital role in ensuring food security and supporting economic development; however, farmers continue to face challenges such as inappropriate crop selec-tion, unpredictable weather conditions, excessive fertilizer usage, fluctuating market prices, and the lack of integrated decision-support systems. To address these challenges, this paper proposes a Smart Agriculture Assistant, an integrated Artificial Intelli-gence (AI)-driven decision support system that combines crop recommendation, crop yield prediction, fertilizer optimization, and market price forecasting within a unified framework. The proposed system utilizes multiple machine learning algorithms, including Random Forest, XGBoost, CatBoost, Long Short-Term Memory (LSTM), and SHAP (SHapley Additive exPlanations), to analyze agricultural and market datasets collected from Kaggle and Agmarknet. The framework processes soil nutrients (N, P, K), pH, temperature, humidity, rainfall, crop characteristics, and historical market price data to generate intelligent recommen-dations and predictive insights for farmers. Data preprocessing techniques such as missing value handling, normalization, feature selection, categorical encoding, and train–test splitting are em-ployed to improve model performance and prediction reliability. Experimental evaluation demonstrates that the proposed hybrid framework achieves an accuracy of 99%, with 98% precision, 98% recall, and 98% F1-score, outperforming conventional machine learning models in agricultural decision-making tasks. The proposed system enables farmers to optimize crop selec-tion, improve fertilizer utilization, enhance crop productivity, and identify profitable market opportunities, thereby promoting sustainable, data-driven, and precision agriculture practices.