Comprehensive evaluation framework for compost maturity with biochar amendment

A predictive framework combining machine learning and weighting techniques was established to resolve inconsistencies in maturity evaluation of biochar-amended composting. The results indicated that the nonlinear model showed superior compost maturity prediction accuracy. Specifically, Gradient boosting (GB), extra trees (ET, used for both GI and NO₃^--N), and extreme gradient boosting (XGB) achieved the highest R² values for C/N ratio (0.84), GI (0.64), NO₃^--N (0.77), and NH₄⁺-N (0.81), respectively. SHAP analysis identified composting process parameters such as moisture content (MC_P), temperature (TEMP_P), and pH (pH_P) as key drivers of enzymatic activity and microbial succession, significantly affecting maturity. The model's applicability and predictive capability were validated through cosine similarity and real-world composting experiments. An integrated maturity score, based on weighted predicted indicators, highlighted GI as the most influential factor (47.62 %). This framework enhances intelligent composting, safer agriculture, and environmental management through predictive accuracy and systematic evaluation.