FA-100006 / Evaluation metric aggregation / Open access
Ignored labels remain in the accuracy denominator · case 01
Ignored labels remain in the accuracy denominator.
ROOT CAUSE
Ignored examples are removed from the numerator but still dilute the score.
THE FAILURE
Ignored examples are removed from the numerator but still dilute the score.
Unsuccessful approach: Filtering out label zero also removes valid zero-class examples.
Case contract
Return accuracy excluding positions whose true label equals ignore; no eligible positions yields 0.
Why this case matters
A small explicit evaluation contract makes aggregation behavior reproducible across labeled examples.
1 / The failure
Exit 1"""Failure Map reference implementation. Python standard library only."""
import json
N = 1
observations = []
def solve(truth, pred, ignore):
valid=[(a,b) for a,b in zip(truth,pred) if a!=ignore]; return sum(a==b for a,b in valid)/len(truth) if truth else 0
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('ignored incorrect',solve([0,1,1],[1,1,0],0),.5)
check('all ignored',solve([0]*N,[1]*N,0),0)
check('zero is ordinary',solve([0,1],[0,0],9),.5)
check('empty',solve([],[],0),0)
check('all eligible correct',solve([N]*N,[N]*N,0),1)
check('single eligible miss',solve([0,N],[1,N+1],0),0)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)
| Boundary fixture | Actual | Expected | Outcome |
|---|---|---|---|
| ignored incorrect | 0.3333333333333333 | 0.5 | Failed |
| all ignored | 0.0 | 0 | Passed |
| zero is ordinary | 0.5 | 0.5 | Passed |
| empty | 0 | 0 | Passed |
| all eligible correct | 1.0 | 1 | Passed |
| single eligible miss | 0.0 | 0 | Passed |
SHA-256 / 5697c3e833e001f7d353c30adf564504feff7a0ca31bdfb2ee7c7cd6b8c531cb
2 / The unsuccessful fix
Exit 1"""Failure Map reference implementation. Python standard library only."""
import json
N = 1
observations = []
def solve(truth, pred, ignore):
valid=[(a,b) for a,b in zip(truth,pred) if a!=ignore and a!=0]; return sum(a==b for a,b in valid)/len(valid) if valid else 0
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('ignored incorrect',solve([0,1,1],[1,1,0],0),.5)
check('all ignored',solve([0]*N,[1]*N,0),0)
check('zero is ordinary',solve([0,1],[0,0],9),.5)
check('empty',solve([],[],0),0)
check('all eligible correct',solve([N]*N,[N]*N,0),1)
check('single eligible miss',solve([0,N],[1,N+1],0),0)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)
| Boundary fixture | Actual | Expected | Outcome |
|---|---|---|---|
| ignored incorrect | 0.5 | 0.5 | Passed |
| all ignored | 0 | 0 | Passed |
| zero is ordinary | 0.0 | 0.5 | Failed |
| empty | 0 | 0 | Passed |
| all eligible correct | 1.0 | 1 | Passed |
| single eligible miss | 0.0 | 0 | Passed |
SHA-256 / db3581cbfea717e631780a67d4ebd583719b23b3431d1eef1b9565e500c391ba
HELD IN THE MEMBER ARCHIVE
The verified repair and its recorded checks are member-only.
This mechanism has 6 recorded checks per implementation. The open-access tier publishes the failure and the unsuccessful fix; the repaired source that passes every check, and the observations that prove it, are available to members.
Every case sharing this mechanism uses the same contract and the same repair, so this one record is held back for all of them.
Member access is invitation-based. Sign in with your invited account to inspect the repair.
Sign in to the archive ↗Verification & scope
Inputs are finite toy labels and scores; this defines no benchmark protocol, production metric library, or model quality claim. This reproducer isolates one failure mechanism. Results cover the supplied fixtures. Variants within a family share a test contract and should remain grouped when constructing evaluation splits. Related mechanisms with a shared evaluation_group must also remain together; these controlled models are not independent production incidents.
Observations recorded using Python 3.12.14 at 2026-09-30T16:48:30.810537+00:00.
Case digest / 77de72af3ba75fa8af87a187015f4d8cfc0a189d4bd08225e4e9947b9f4e47c3