{"id":"W4392757155","doi":"10.1080/10920277.2023.2279782","title":"A Flexible Hierarchical Insurance Claims Model with Gradient Boosting and Copulas","year":2024,"lang":"en","type":"article","venue":"North American Actuarial Journal","topic":"Insurance and Financial Risk Management","field":"Economics, Econometrics and Finance","cited_by":8,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université Laval","funders":"","keywords":"Copula (linguistics); Boosting (machine learning); Gradient boosting; Econometrics; Computer science; Economics; Artificial intelligence","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003504946,0.0002225783,0.0004570885,0.0003114552,0.0003072238,0.0003921457,0.0001949893,0.00003738999,0.0000250212],"category_scores_gemma":[0.00004151007,0.0002002314,0.0001039451,0.0005432692,0.0002530321,0.0003729611,0.00005792958,0.0006177827,0.00007140524],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001201159,"about_ca_system_score_gemma":0.00008639909,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000203979,"about_ca_topic_score_gemma":0.0001425245,"domain_scores_codex":[0.998403,0.00001618367,0.0005553453,0.0004306338,0.0001078009,0.0004870407],"domain_scores_gemma":[0.999262,0.00004401792,0.0002813942,0.0001778419,0.00003378199,0.000200906],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0006438162,0.0001588291,0.5124783,0.0001007119,0.0003115355,0.0004098458,0.002823577,0.009678169,0.000009741998,0.1236292,0.002475726,0.3472805],"study_design_scores_gemma":[0.002287479,0.001765213,0.7523789,0.0002472617,0.00006448806,0.000542079,0.0002344989,0.0921443,0.00001906832,0.04424531,0.1046122,0.001459159],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9258173,0.001096878,0.06660754,0.0008088783,0.0006008772,0.0001759594,0.00009067688,0.00007793096,0.00472399],"genre_scores_gemma":[0.9935434,0.001273228,0.003793199,0.0004486916,0.0005935465,0.00001412432,0.000006135214,0.00003928323,0.0002884049],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3458214,"threshold_uncertainty_score":0.81652,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01965093821642587,"score_gpt":0.2227416359532634,"score_spread":0.2030906977368375,"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."}}