Convenient Classification of Phyllostachys heterocycla cv. pubescens Properties Based on Growth Traits and Machine Learning Methods

Bamboo is widely regarded as an eco-friendly construction material due to its rapid growth cycle, strong adaptability, high productivity, and excellent mechanical properties, positioning it as one of the most promising forest-based alternatives to wood. Implementing a property-based classification system that correlat es growth traits with material properties is fundamentally important for a dvancing the rational utilization of bamboo resources. This study employed K-Means, decision tree algorithms, and regression modeling of LASSO and SVR in machine learning to categorize bamboo properties using four key growth traits: age, diameter of breast height, height, and wall thickness. The results demonstrated that growth traits serving as decision tree root nodes and intermediate nodes significantly contribute to bamboo property classification, can reflect the integrated influence of growth traits on material properties, and enable effective and convenient property classification. The material property regression model exhibited rather low fitness, indicating that growth traits alone were insufficient for developing high-accuracy predictive equations.