Machine Learning in Environmental Research: Common Pitfalls and Best Practices

Machine learning (ML) is increasinglyused in environmentalresearchto process large data sets and decipher complex relationships betweensystem variables. However, due to the lack of familiarity and methodologicalrigor, inadequate ML studies may lead to spurious conclusions. Inthis study, we synthesized literature analysis with our own experienceand provided a tutorial-like compilation of common pitfalls alongwith best practice guidelines for environmental ML research. We identifiedmore than 30 key items and provided evidence-based data analysis basedon 148 highly cited research articles to exhibit the misconceptionsof terminologies, proper sample size and feature size, data enrichmentand feature selection, randomness assessment, data leakage management,data splitting, method selection and comparison, model optimizationand evaluation, and model explainability and causality. By analyzinggood examples on supervised learning and reference modeling paradigms,we hope to help researchers adopt more rigorous data preprocessingand model development standards for more accurate, robust, and practicablemodel uses in environmental research and applications.