terazi - AI Fairness for Doubly Imbalanced Data


« back to home

User dataset

vehicle_insurance.csv

Favourable label:

"FraudFound_P" = 0

Unfavourable label:

"FraudFound_P" = 1

Privileged group:

"Sex" = "Female"

Unprivileged group:

"Sex" = "Male"

Data statistics:



Demo

Show the pareto graph using with search depth:





\(\alpha:\)

\(\beta:\)

\(\gamma:\)




Optimization metrics
DI Ratio #
MCC #
Combined Loss #