Agricultural Crop Recommendation System for Farmers Using Machine Learning Techniques

The socioeconomic fabric of India is heavily dependent on agriculture. Failure on the part of farmers to select the crop that is best suited to the land through traditional and non-scientific methods. The CNN, RF, and KNN-based agriculture crop recommendation system is profitable and efficient. It was implemented with the intention of reducing resource wastage, increasing crop health, and mitigating the risks posed by climate variability. Preprocessing steps like data collection, data cleaning, and feature extraction were used in this study because they are most relevant to the chosen analysis. In relation to the problem domain, numerous supervised and unsupervised machine learning algorithms are investigated. The implemented system suggests crops to farmers based on the type of soil, region, and location. The chosen models are trained and hyperparameter tuned to improve efficiency and accuracy using techniques like Grid Search, Random Search, and Bayesian Optimization. Metrics like accuracy, precision, recall, and the F1-score are used in evaluation and validation to ensure the model's robustness. The optimized model is then incorporated into existing systems or implemented in a web- based application for real-world usability. This framework's systematic, scalable, and efficient approach to building high- performing predictive models. In this research trained CNN model on 100 epochs and achieved the 95.25% accuracy.