Predicting Water Retention in Vicia faba Under Salt Stress Conditions Using Machine Learning

Abstract

This study explores the potential of artificial intelligence (AI) techniques to predict Relative Water Content (RWC) in plants subjected to salt stress. A Random Forest regression model was trained using a dataset containing both morphological and biochemical traits, such as chlorophyll content, dry biomass, proline concentration, and total sugar levels. Data preprocessing addressed format inconsistencies and missing values, and feature importance analysis was conducted to assess the influence of each variable. The model showed a Mean Absolute Error (MAE) of 3.88 on the test set and a cross-validated MAE of 5.18, demonstrating reliable performance despite biological variability. These results align with known physiological relationships, validating the model's interpretability and usefulness in plant stress diagnostics. This work highlights AI's potential to enhance decision-making in agronomy and improve stress management strategies.