Identifying causal pathways and risk-decision rules for nitrous oxide emission hot moments in wastewater treatment plants using probabilistic causal machine learning

The code package includes scripts for: Cleaneddata – data preprocessing and preparation of the cleaned dataset used for subsequent analyses; Temporal characteristics – analysis of the temporal characteristics and patterns of N₂O emissions and associated operational variables; XGBoost – XGBoost model development and analysis of N₂O emissions, including model training and evaluation; Cohort SHAP – cohort-based SHAP analysis for interpreting feature contributions under different operational conditions; Copula – copula-based joint probability analysis of key operational-variable combinations and N₂O emission risks; LiNGAM – exploratory causal analysis using the LiNGAM framework to identify potential causal relationships among operational variables and N₂O emissions.

These scripts provide the computational workflow underlying the results reported in the manuscript and are provided to facilitate transparency, reproducibility, and independent verification of the analyses.