Prediction of Available Phosphorus Formation and Analysis of Key Influencing Factors during Organic Waste Composting Using Stacking Ensemble Learning
Organic waste is a potential phosphorus reservoir, and understanding the dynamics of available phosphorus (AP) during its resource utilization is critical for efficient phosphorus recovery. Composting, a key route for organic waste valorization, involves complex transformations of phosphorus alongside organic matter degradation and humification. However, the long duration and high cost of composting experiments, coupled with multifactorial influences, hinder efficient elucidation of AP dynamics via conventional methods. This study compiled data from 33 publications, constructing a dataset of 647 samples. Data preprocessing included iterative imputation, one-hot encoding, and standardization. A stacking ensemble learning model was developed to predict AP generation during composting. The optimal ensemble comprised XGBoost and SVR as base learners and ElasticNet as the meta-learner, achieving R² values of 0.954 and 0.928 on training and test sets, respectively, with low overall error. SHAP analysis revealed that key factors influencing AP content, in descending order of importance, were feedstock type, bulking agent type, turning interval, pH, electrical conductivity (EC), and C/N ratio. Notably, livestock manure as feedstock and straw-based bulking agents contributed positively to AP predictions. Partial dependence plots indicated that lower pH and C/N ratios generally favored AP accumulation throughout composting. During the initial stage, higher moisture content and lower EC enhanced AP; in the thermophilic phase, higher temperatures corresponded to higher AP; and during cooling and maturation, maintaining moisture below 48% and C/N below 14, while extending composting beyond 43 days, promoted AP accumulation. This study demonstrates accurate AP prediction via stacking ensemble learning and identifies critical factors, offering support for optimizing phosphorus management in composting engineering.