A Comparative Analysis of Machine Learning Models for Crop Classification Using Soil Nutrients and Environmental Data
Abstract
Purpose: Agriculture is essential to the growth of the world economy. Food availability today largely depends on whether crops have been harvested properly. The machine learning methods described in this research use soil nutrient and environmental data to classify crops. Methodology: This study focuses on using only current soil and weather data to classify suitable crops. Several machine learning algorithms, such as light gradient boosting, decision trees, random forests, and logistic regression, are applied to evaluate crop classification models on soil and environmental data. Findings: To calculate accuracy, precision, recall, and f1-score, this research used the available data from Kaggle. The decision tree and random forest algorithms achieve an approximately 99% accuracy by using the cross-validation method under the applied evaluation setting. Originality: A comparison with other existing models reveals that the decision tree and random forest algorithms get higher results among the applied models. These models achieved the highest accuracy because they effectively capture complex, nonlinear relationships between soil nutrients and environmental factors, and reduce overfitting. Practical implications: The findings of this work may support future research and decision-support tools for crop classification models.