Unverified50% confidenceOpinionExact time
How the chronologically split BGL result (0.870) can be fairly compared with conventionally reported random-split results still requires community consensus.
1
Sources
50%
Confidence
Long-term
Relevance
11/10/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSystems with the same overall accuracy may produce markedly different error distributions across two population groups, for example frequently wrongly rejecting applicants in group A who would have repaid on time, and more often wrongly approving applicants in group B who ultimately defaulted.64% similarUnverifiedFairness is measured by Equalized Odds, a metric that requires the model to maintain consistent true positive rates and false positive rates across different demographic groups.63% similarUnverifiedThe fairness standard ProPublica used is "group balance in predictive error rates," while the company that developed COMPAS countered that the system is calibrated on "predictive accuracy" for both groups, i.e., the actual recidivism rate corresponding to a high risk score is roughly the same across races.62% similarUnverifiedOn the consensus reference standard formed after review, another annotator's agreement with that consensus on 200 randomly sampled tweets was 60.5% (κ=0.517).62% similarUnverifiedThe experiments designed control groups for persistence and dominance results, keeping the accuracy of random guessing at around 0.27.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1154294API
curl https://kongchang.com/api/v1/knowledge/claims/1154294MCP
get_claim(id=1154294)