Reproducible workflow for explainable machine learning analysis of biochemical methane potential

Résumé fourni par la source

Initial public release This release provides the reproducibility materials supporting the manuscript "An Explainable Machine Learning Framework for Hypothesis Generation in Biochemical Methane Potential Prediction." The repository includes: the processed analysis dataset and source-provenance records; nested cross-validation, prediction-error, SHAP, and sensitivity-analysis code; row-level out-of-fold predictions and supporting numerical outputs; manuscript and supplementary figures in multiple formats; the journal-facing row-level supplementary dataset and data dictionary; environment specifications and reproducibility instructions. The primary nested Random Forest performance was R² = 0.530, RMSE = 44.5 Nm³ CH₄/t DM, and MAE = 33.6 Nm³ CH₄/t DM. This release corresponds to commit 4f23339ff5951a9a7ce7eb7d060b30770cd1fe81. Final validated archive for manuscript revision. SHA-256: 99C4DF7231D4CFF941499C4ED3A3C2F6F82C5E010DCC29896B71ED75E16253CF. Please use the repository's CITATION.cff file for citation metadata. License: MIT.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.