The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.
aigov_drift() audits decision drift between two
periods. It decomposes the change in each decision rate into case-mix
and rule-change components, and, when a decision unit is supplied
(recruiter, office, judge, court), into case-mix, allocation and
within-unit rule-change components. The components add to the raw change
exactly. It also reports net drift and absolute (total-variation) drift
over a fixed reference case mix, with overlap trimming, cluster
bootstrap intervals, a bias-corrected absolute drift and alerts that
distinguish net drift from masked drift (offsetting changes that leave
aggregate rates unchanged). Alerts use interval nulls: an alert requires
the confidence interval to lie beyond the practical tolerance, so
changes plausibly within tolerance never trigger one.
net_p_tol gives the matching one-sided p-value, and
aigov_monitor() adjusts it across comparisons.decision argument.aigov_monitor() runs prespecified comparisons across
several periods (against the baseline or the previous period) with
multiplicity adjustment.aigov_simulate_drift() simulates decision data from
known mechanisms (stable, case mix, rule change, offsetting group or
case changes, unrecorded characteristic, reallocation across units,
offsetting unit changes, category shift) and returns the true drift
targets.hiring_panel - three periods, 24 recruiters,
three-category decision, with known drift mechanisms.aigov_drift() or aigov_monitor()
now fills NIST AI RMF item MEASURE_3_1 (ongoing monitoring for bias
drift) in aigov_audit_nist(). MANAGE_4_1 still requires a
human review and is not set automatically.aigov_checklist(gov, "EU_AI_Act") the item shows “Evidence
available” once a drift audit has been run.aigov_report() includes decision-drift and monitoring
sections when those modules have been run, and reports the installed
package version.utils is now declared in Imports; nnet is
a new import.aigov_build() — construct governance audit object from
employment dataaigov_scope() — identify applicable frameworks by
domain and jurisdictionaigov_adverse_impact() — EEOC 4/5ths adverse impact
rule with Z and Fisher testsaigov_audit_nyc() — NYC Local Law 144 impact ratio
table and procedural checklistaigov_audit_nist() — NIST AI RMF 1.0
GOVERN/MAP/MEASURE/MANAGE scoringaigov_classify() — EU AI Act risk tier and NIST risk
tier classificationaigov_checklist() — display checklist items for any
supported frameworkaigov_report() — generate self-contained HTML audit
reporthiring_sim — synthetic employment screening dataset
(500 applicants, 5 race/ethnicity groups)These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.