Published January 1, 2026
| Version v1
Journal article
Open
A Comparative Analysis of Machine-Learning Models for Plant Acoustic Stress Classification
Authors/Creators
Description
In this paper, the authors develop a comparative framework for acoustic stress analysis of plants, which is based on a primary sensing modality, acoustic observations, and a secondary one, environmental measurements. The provided evidence includes a dataset containing 3,991 environmental records labeled with 10 classes and a performance comparison table for six different machine learning models: Support Vector Machine (SVM), Random Forest (RF), CatBoost, XGBoost, Improved XGBoost, and Stacking Classifier. The reported accuracies are 79.97%, 82.73%, 82.10%, 82.35%, 82.23%, and 82.60%, respectively. Due to the fact that the attached tabular data does not include the actual acoustic descriptors but rather environmental variables, these numerical results are explicitly reported as an environmental/contextual computational baseline, and are not considered to be proven acoustic-only performance. The manuscript thus offers a leakage-aware, step-by-step approach for gathering acoustic signals, reducing environmental noise, extracting acoustic features, separating acoustic and contextual information, training the six classifiers, and testing plant level generalization. The framework is designed to enable a subsequent controlled acoustic experiment with plant-level splits to independently label and assess the Healthy, Unstressed and Stressed states. The final manuscript keeps the provided results and also keeps a good separation between evidence measured and the proposed acoustic experiment.
Files
IJSET_V14_issue5_157.pdf
Additional details
Related works
- Has part
- Journal article: https://www.ijset.in/wp-content/uploads/IJSET_V14_issue5_157.pdf (URL)
- Is identical to
- Journal article: https://www.ijset.in/a-comparative-analysis-of-machine-learning-models-for-plant-acoustic-stress-classification/ (URL)