Home · Analysis

17 analysis units, one page each

These are the things a comparison has to answer. They are grouped as a programme would run them, from the object and the decision through accuracy and equality to the report and the re-run. Each unit is our statement of the public source named on it; the DIS body is not public.

The unit codes are ours. They are not ISO clause numbers, and no page here claims that they are. Where a unit rests on the public working draft of the benchmarking part, the page says so and names the draft clause.

Scope the comparison

Code	Unit	Public source
P3-01	Name the object you are comparing	WG 5 N3928, clause 4.1.1
P3-02	Name the decision the comparison feeds	WG 5 N3928, clause 4.1.2
P3-03	Choose the benchmark basis	WG 5 N3928, clauses 4.1.3 and 4.2

Indicators of effectiveness

Code	Unit	Public source
P3-04	State the headline accuracy against an indicator	WG 5 N3928, clause 4.3.1
P3-05	Qualify accuracy with the age buffer	WG 5 N3928, clause 4.3.2
P3-06	Report continuous measures for an estimated age	WG 5 N3928, Annex A
P3-07	Report binary measures at the decision threshold	WG 5 N3928, Annex B
P3-08	Test equality across groups	WG 5 N3928, clause 4.4.1.2 and Annex C

Method and evidence

Code	Unit	Public source
P3-09	Select the test methods and the test crew	WG 5 N3928, clause 4.4.2
P3-10	Set the sample, the sampling frame and the demographics	WG 5 N3928, clause 4.1.2 and Annexes A and B
P3-11	Test presentation attack detection	WG 5 N3928, clause 4.4.3
P3-12	Assess vulnerability and rate the attack	WG 5 N3928, clause 4.4.4
P3-13	Quantify uncertainty and repeatability	WG 5 N3928, clause 4.4.1.4

Report and maintain

Code	Unit	Public source
P3-14	Write the analysis report	WG 5 N3928, clauses 4.3.3 and 4.4
P3-15	Publish the claim and the method	WG 5 N3928, clause 4.3.3
P3-16	Keep the comparison reproducible	WG 5 N3928, clauses 4.4.1.3 and 4.4.2
P3-17	Set the re-run triggers	WG 5 N3928, clause 4.4.1.3