FA-100371 / Feature preprocessing fit scope / Open access

Category vocabulary encoding uses held-out rows while fitting · case 01

Category vocabulary encoding uses held-out rows while fitting.

Verified by executionVariant 1 · 7 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

Fitting reads values outside the designated training rows.

THE FAILURE

Fitting reads values outside the designated training rows.

Unsuccessful approach: Filtering rows after computing the statistic still leaves held-out influence.

Case contract

Fit sorted unique string classes from train and return their zero-based positions for transform values; any unseen string receives sentinel -1.

Why this case matters

A small deterministic preprocessing contract makes the fit boundary and transform behavior directly inspectable.

1 / The failure

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(train, values):
    vocab=sorted(set(train+values)); return [vocab.index(x) if x in vocab else -1 for x in values]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
a='c'+str(N); train=['b',a]; vals=[a,'b','u'+str(N),'']; check('lexical train ids',solve(train,vals),[1,0,-1,-1])
check('heldout category unknown',solve(['a','c'],['b']),[-1])
check('repeat',solve(['z','a','z'],['z','z']),[1,1])
check('empty values',solve(['a'],[]),[])
check('empty fit',solve([],['a']),[-1])
check('empty string unknown',solve(['a'],['']),[-1])
check('known singleton',solve(['x'],['x']),[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
lexical train ids	`[2, 1, 3, 0]`	`[1, 0, -1, -1]`	Failed
heldout category unknown	`[1]`	`[-1]`	Failed
repeat	`[1, 1]`	`[1, 1]`	Passed
empty values	`[]`	`[]`	Passed
empty fit	`[0]`	`[-1]`	Failed
empty string unknown	`[0]`	`[-1]`	Failed
known singleton	`[0]`	`[0]`	Passed

SHA-256 / 710c7c3fa13ea05b33703870f96268860bd32012eedc9aeab85002e51d34bc4e

2 / The unsuccessful fix

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(train, values):
    vocab=sorted(set(train+values)); return [vocab.index(x) for x in values]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
a='c'+str(N); train=['b',a]; vals=[a,'b','u'+str(N),'']; check('lexical train ids',solve(train,vals),[1,0,-1,-1])
check('heldout category unknown',solve(['a','c'],['b']),[-1])
check('repeat',solve(['z','a','z'],['z','z']),[1,1])
check('empty values',solve(['a'],[]),[])
check('empty fit',solve([],['a']),[-1])
check('empty string unknown',solve(['a'],['']),[-1])
check('known singleton',solve(['x'],['x']),[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
lexical train ids	`[2, 1, 3, 0]`	`[1, 0, -1, -1]`	Failed
heldout category unknown	`[1]`	`[-1]`	Failed
repeat	`[1, 1]`	`[1, 1]`	Passed
empty values	`[]`	`[]`	Passed
empty fit	`[0]`	`[-1]`	Failed
empty string unknown	`[0]`	`[-1]`	Failed
known singleton	`[0]`	`[0]`	Passed

SHA-256 / 948aba7cf5d5bb109518236d4d1799ac68b75a8e15ea25d97529399a08c0314d

HELD IN THE MEMBER ARCHIVE

The verified repair and its recorded checks are member-only.

This mechanism has 7 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.

Verification & scope

Toy categorical ordinal encoding with deterministic lexical order and a single unknown sentinel. 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 2026-09-30T16:48:34.921039+00:00.

Case digest / 996095f1da714491522c958419c5fdd6ad53c2981118b3ad705733d2a5724da0