Hybrid Ensemble Approach to Predict Plant Growth for Enhancing Agricultural Productivity

Abstract

Accurate prediction of plant growth milestones is essential for optimizing agricultural practices and enhancing greenhouse management. This study addresses the challenge of classifying plant growth stages by leveraging environmental and management factors, including soil type, sunlight exposure, watering frequency, fertilizer type, temperature, and humidity. We utilized a comprehensive dataset encompassing these variables to develop a robust predictive model. The methodology involved meticulous data pre-processing steps, including handling missing values, encoding categorical variables, and scaling numerical features to prepare the data for analysis. To advance the state-of-the-art in plant growth prediction, we proposed a novel hybrid ensemble model that integrates multiple machine learning algorithms—specifically, Random Forest, Gradient Boosting, and a Neural Network—and employs a meta-learner, Logistic Regression, to synthesize their predictions. This ensemble approach was designed to harness the strengths of each individual model, thereby enhancing overall predictive performance. We conducted a thorough evaluation of the proposed hybrid model against individual baseline models using metrics such as accuracy, precision, recall, and F1-score. Our results demonstrate that the hybrid ensemble model significantly outperforms the baseline models, achieving an accuracy of 89.1%, compared to 85.2% for Random Forest, 87.4% for Gradient Boosting, and 86.8% for the Neural Network. Additionally, the hybrid model excelled in other evaluation metrics, including precision (88.7%), recall (89.5%), and F1-score (89.1%), showcasing its superior performance. Feature importance analysis revealed that factors such as sunlight exposure and watering frequency are critical determinants of plant growth milestones. This research contributes to the field by presenting a novel, data-driven approach that enhances the accuracy of plant growth predictions, thereby offering valuable insights for improving agricultural productivity and sustainability.