{"id":"W4307622887","doi":"10.3233/sji-220082","title":"Predicting the quality and evaluating the use of administrative data for the 2021 Canadian Census of Population","year":2022,"lang":"en","type":"article","venue":"Statistical Journal of the IAOS","topic":"Census and Population Estimation","field":"Mathematics","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Statistics Canada","funders":"","keywords":"Census; Imputation (statistics); Data quality; Population; Pandemic; Geography; Missing data; Coronavirus disease 2019 (COVID-19); Computer science; Statistics; Environmental health; Medicine; Operations management; Economics; Mathematics","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03737157,0.0008346984,0.0006274448,0.002368653,0.001473671,0.002827278,0.002581809,0.0007182471,0.001154842],"category_scores_gemma":[0.15645,0.0005490339,0.0006314309,0.005704874,0.001126537,0.001481203,0.001671317,0.001146195,0.0002975959],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0176502,"about_ca_system_score_gemma":0.03148814,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.8968762,"about_ca_topic_score_gemma":0.9016129,"domain_scores_codex":[0.9815709,0.01113934,0.0006277199,0.0009092203,0.00464881,0.001104016],"domain_scores_gemma":[0.9068618,0.0551471,0.005337926,0.007204795,0.02323486,0.002213436],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0004775427,0.000337325,0.5369073,0.0002337074,0.0003707438,0.0001055448,0.0009433432,0.3322975,0.0004769819,0.01525878,0.01286315,0.09972802],"study_design_scores_gemma":[0.0001192296,0.0002056484,0.1689581,0.000170603,0.0001425121,0.00003108957,0.00125866,0.8136569,0.001407432,0.004196182,0.009755411,0.00009825268],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.818426,0.001365168,0.128323,0.009503363,0.000168216,0.001394068,0.02281917,0.001086232,0.01691484],"genre_scores_gemma":[0.8793241,0.0005110705,0.1065919,0.0002833756,0.00003233088,0.0003275627,0.01186673,0.00007504047,0.0009878471],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1031238,"threshold_uncertainty_score":0.2074621,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.5978813152470713,"score_gpt":0.523511196845447,"score_spread":0.07437011840162433,"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."}}