Analysis of soil suitability for agricultural needs using machine learning methods

Abstract

This study explores the application of machine learning methods to assess soil suitability for agricultural purposes, focusing on identifying and analysing key factors that influence soil productivity under drought conditions. A classification model was developed based on data from diverse U.S. regions, which included critical soil parameters such as root condition, nutrient availability, soil toxicity, and oxygen accessibility for plant roots. Correlation analysis identified the most significant factors impacting soil suitability, with root condition and nutrient availability emerging as primary determinants. The model achieved a high accuracy of 98.81%, demonstrating its effectiveness in predicting soil suitability across varying environmental conditions. This research highlights the value of machine learning in optimizing agricultural practices by enabling data-driven soil assessment. Moreover, the study addresses the importance of accessible and cost-effective data collection methods for scaling the model to different regions. Despite the model’s high accuracy, limitations related to data specificity and regional variability were observed, indicating areas for future improvement. Expanding the model with additional climatic and geographic data could enhance its generalizability and applicability in diverse agricultural settings. Overall, this study provides insights into the potential of machine learning to support sustainable agriculture and efficient land management.