{"id":"W3154817805","doi":"10.1016/j.advwatres.2021.103920","title":"A workflow to address pitfalls and challenges in applying machine learning models to hydrology","year":2021,"lang":"en","type":"article","venue":"Advances in Water Resources","topic":"Hydrological Forecasting Using AI","field":"Environmental Science","cited_by":53,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Alberta","funders":"Mitacs","keywords":"Computer science; Streamflow; Hydrological modelling; Machine learning; Feature selection; Dimensionality reduction; Artificial intelligence; Data mining; Model selection; Test set; Cross-validation; Data pre-processing; Workflow; Drainage basin","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008146044,0.002037033,0.001189472,0.004312879,0.00214263,0.007851234,0.002816234,0.002470414,0.02700723],"category_scores_gemma":[0.03184862,0.001182059,0.001897821,0.002774486,0.0010427,0.003738873,0.004758927,0.003601155,0.02118161],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001275448,"about_ca_system_score_gemma":0.008903331,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009935519,"about_ca_topic_score_gemma":0.009404995,"domain_scores_codex":[0.9973897,0.0006185772,0.000484388,0.0003674442,0.0009493412,0.0001905396],"domain_scores_gemma":[0.9777899,0.007876375,0.0006646874,0.006503987,0.005821369,0.001343801],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0006474522,0.0006923589,0.00568034,0.0008354047,0.0002046925,0.002112498,0.002277452,0.02063601,0.01722905,0.05972885,0.3355473,0.5544086],"study_design_scores_gemma":[0.0004254546,0.0001395611,0.002680645,0.0006103369,0.0001201923,0.001012711,0.0008226926,0.2147315,0.04872215,0.1769996,0.5534012,0.0003339029],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.002567365,0.00009661113,0.8900575,0.002311812,0.0005039239,0.0008040515,0.004208315,0.09254354,0.006906824],"genre_scores_gemma":[0.01927231,0.0002386999,0.9530826,0.0007382476,0.0001523521,0.0008978156,0.007334742,0.008298133,0.009985128],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.02700723,"threshold_uncertainty_score":0.09034818,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04900749370254814,"score_gpt":0.2547765371059854,"score_spread":0.2057690434034372,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}