{"id":"W4385462613","doi":"10.1038/s41559-023-02162-1","title":"Data leakage jeopardizes ecological applications of machine learning","year":2023,"lang":"en","type":"article","venue":"Nature Ecology & Evolution","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":29,"is_retracted":false,"has_abstract":false,"ca_institutions":"Sciencetech (Canada); University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Computer science; Machine learning; Leakage (economics); Artificial intelligence; Ecology; Data science; Biology","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":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0004138166,0.00009093173,0.0001431213,0.00005211793,0.0001994992,0.000006363842,0.0004111853,0.0003277944,0.01689137],"category_scores_gemma":[0.0002541809,0.00008331244,0.00003991049,0.0006199992,0.0002164868,0.0001137681,0.0005816253,0.0004311741,0.002778297],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003289052,"about_ca_system_score_gemma":0.00001066781,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003398276,"about_ca_topic_score_gemma":0.001845853,"domain_scores_codex":[0.9989803,0.00008535382,0.0001759753,0.0003369828,0.0001715493,0.0002498594],"domain_scores_gemma":[0.9993688,0.0001240702,0.000105068,0.0003360569,0.00001562194,0.00005039285],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00002416953,0.0002110504,0.9101959,0.00001313639,0.00002100395,0.000004219921,0.00003520357,0.0004164506,0.008305241,0.005614193,0.07403538,0.00112406],"study_design_scores_gemma":[0.0001751852,0.00004944163,0.8810509,9.645197e-7,0.00001518932,0.00000485896,0.0001903562,0.002188795,0.0001105297,0.0003648676,0.1157741,0.00007480564],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9794171,0.000206406,0.0005971241,0.001508764,0.0003110725,0.0003885126,0.0005738162,0.0002974199,0.01669973],"genre_scores_gemma":[0.9969447,0.0001082093,0.00007418785,0.00009471448,0.00004124383,0.00004995369,0.001856416,0.000006769902,0.0008237988],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.04173876,"threshold_uncertainty_score":0.9979982,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02240264507853213,"score_gpt":0.2840667524177297,"score_spread":0.2616641073391976,"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."}}