Accessing the Efficiency of Machine learning Models for Crop Categorization Based on Regional Attributes in India

This comprehensive model comparison demonstrates assessment of various machine learning methods designed for identifying the different crops grown in India and categorizing them, tailoring the classification to the specific attributes of each State. Some drawbacks observed in previously implemented systems and studies arise from their singular focus on a specific parameter—be it moisture, temperature, soil pH, etc.—for predicting the adequacy of crop growth. This project focuses on a multitude of factors like soil quality indicated by parameters such as pH, Nitrogen (N), Phosphorus (P), Potassium (K) content, in addition to temperature and rainfall. The primary models utilized and compared are Decision Tree, K-Nearest Neighbors, Random Forest, Extreme Gradient Booster (XGBoost), and Naïve Bayes. K Nearest Neighbors emerges as the best model with an f1-score of 97% and a accuracy of 94%. The research investigates the connection between model complexity and accuracy, demonstrating how the choice of a particular model influences practical outcomes in crop prediction. The models underwent training and validation using a dataset structured in CSV format, comprising a total of 1,000,000 crop records. This training was conducted within a supervised learning framework. Comparison metrics for these models were derived from standardized parameters, encompassing losses, accuracy, and other pertinent metrics