{"id":"W2912895282","doi":"10.1016/j.neucom.2018.11.100","title":"Pre-processing approaches for imbalanced distributions in regression","year":2019,"lang":"en","type":"article","venue":"Neurocomputing","topic":"Imbalanced Data Classification Techniques","field":"Computer Science","cited_by":130,"is_retracted":false,"has_abstract":false,"ca_institutions":"Dalhousie University","funders":"European Regional Development Fund; Fundação para a Ciência e a Tecnologia","keywords":"Computer science; Regression; Machine learning; Variable (mathematics); Relevance (law); Regression analysis; Artificial intelligence; Context (archaeology); Feature selection; Set (abstract data type); Data mining; Mathematics; Statistics","routes":{"ca_aff":true,"ca_fund":false,"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":[0.006765726,0.001689174,0.002104245,0.00272383,0.001422967,0.002539647,0.0027021,0.001419364,0.007544813],"category_scores_gemma":[0.01728646,0.0008678383,0.002001711,0.003186472,0.0009122516,0.002798201,0.002581283,0.005103996,0.004465091],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007177599,"about_ca_system_score_gemma":0.002155863,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00248787,"about_ca_topic_score_gemma":0.004555999,"domain_scores_codex":[0.9967638,0.0007699267,0.0003820173,0.0005956788,0.001197802,0.0002907174],"domain_scores_gemma":[0.9902263,0.004381272,0.0005362037,0.001850899,0.002780992,0.0002243567],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004790005,0.0004237845,0.002719666,0.0003663182,0.0001974787,0.0001692327,0.0002579314,0.06144957,0.01449575,0.01675215,0.01055199,0.8921373],"study_design_scores_gemma":[0.00005288128,0.0001845742,0.002600488,0.00007893841,0.0001025256,0.0001711487,0.0001622221,0.9387017,0.0180889,0.02725287,0.01256123,0.00004258295],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.005711281,0.0004913274,0.9913619,0.0002453541,0.0001721718,0.0001158313,0.0001588023,0.001099316,0.0006439079],"genre_scores_gemma":[0.1206556,0.0009245244,0.8690892,0.0003057187,0.0008034613,0.0005035522,0.001785739,0.0005818388,0.005350336],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007544813,"threshold_uncertainty_score":0.03578097,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03953302892903467,"score_gpt":0.2872695688861338,"score_spread":0.2477365399570991,"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."}}