{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004079821,0.0001639461,0.0002517575,0.00009053704,0.00008259921,0.00003072279,0.0002098429,0.00007625598,0.0001274415],"category_scores_gemma":[0.00008496742,0.0001272079,0.00001981177,0.0002392832,0.00009060694,0.0002088962,0.0006559495,0.0002533841,0.00007847558],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006014353,"about_ca_system_score_gemma":0.000001175561,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001729779,"about_ca_topic_score_gemma":0.002715463,"domain_scores_codex":[0.9982772,0.000180562,0.0002328614,0.0006091624,0.000187169,0.0005129796],"domain_scores_gemma":[0.9995678,0.0001015251,0.00002571197,0.000174055,0.000003868354,0.0001270374],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00004529956,0.00003683896,0.01785263,0.0000148264,0.000001866633,0.0001089062,0.007638472,0.8796421,0.002657084,0.00002102312,9.230898e-7,0.09198008],"study_design_scores_gemma":[0.001391523,0.000695947,0.008546581,0.0005015859,0.00001292371,0.0001305402,0.0005468126,0.1891574,0.007237752,0.02561986,0.7648783,0.001280788],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9888526,0.004435066,0.0001623051,0.001792339,0.00004132109,0.0001981801,8.583535e-7,0.00004383183,0.004473498],"genre_scores_gemma":[0.9936545,0.001001038,0.004520494,0.000545645,0.00002289187,0.00008373606,0.000002370598,0.00001765262,0.0001516272],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7648773,"threshold_uncertainty_score":0.5187387,"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."}}