{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.01970065,0.0006093018,0.00093276,0.001917023,0.001323384,0.004433083,0.00232565,0.003444264,0.004488702],"category_scores_gemma":[0.1814016,0.0007386052,0.000782354,0.002810122,0.003103238,0.009383832,0.005176748,0.004214477,0.002017167],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001454491,"about_ca_system_score_gemma":0.002679149,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001873738,"about_ca_topic_score_gemma":0.001235817,"domain_scores_codex":[0.9894422,0.004366964,0.0008985403,0.001071756,0.003666033,0.0005544376],"domain_scores_gemma":[0.7908471,0.1535116,0.007253328,0.03362713,0.01317199,0.001588787],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.002970867,0.0006099539,0.06891713,0.0009118924,0.0004650133,0.0009825779,0.001437656,0.06465524,0.02216372,0.1165399,0.03986564,0.6804805],"study_design_scores_gemma":[0.0002014064,0.0003550871,0.01064516,0.0003663511,0.0002577221,0.001654126,0.0006028881,0.4481373,0.04932541,0.4577118,0.03059816,0.0001445684],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.412867,0.006136165,0.4758373,0.05946236,0.002796213,0.0002423116,0.002171797,0.01285371,0.0276332],"genre_scores_gemma":[0.9577242,0.0008047823,0.03257241,0.004340496,0.0005453943,0.00006325275,0.0005009572,0.0007792336,0.002669293],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9802994,"threshold_uncertainty_score":0.1041882,"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."}}