Compliance dossier — training data (AI Act, Article 10)

Automatically generated skeleton from the dataset analysis. The "to be completed by the expert" blocks require human review (provenance, compliance judgment, mitigation measures).

1. Identity & purpose

System / model: _____
Dossier version: _____
Analysis date: _____
Dataset: src
Intended purpose: _____

_To be completed by the expert — intended purpose and what the data is meant to represent_

2. Provenance & legal basis

Origin of each source, licences/contracts, and — for personal data — initial purpose and GDPR legal basis.

_To be completed by the expert — provenance per source (separate flow §16), licences, GDPR legal basis_

3. Composition _(filled automatically)_

Volume: 32561 rows, 15 columns.
Variables:

Column	Type
`age`	BIGINT
`workclass`	VARCHAR
`fnlwgt`	BIGINT
`education`	VARCHAR
`education_num`	BIGINT
`marital_status`	VARCHAR
`occupation`	VARCHAR
`relationship`	VARCHAR
`race`	VARCHAR
`sex`	VARCHAR
`capital_gain`	BIGINT
`capital_loss`	BIGINT
`hours_per_week`	BIGINT
`native_country`	VARCHAR
`income`	VARCHAR

Declared sensitive attributes: age, workclass, education, marital_status, occupation, relationship, race, sex.

_To be completed by the expert — geographic / contextual / behavioural scope_

4. Preparation _(automatic observations)_

Duplicates: 24 duplicate row(s) → deduplication step to be traced.

_To be completed by the expert — transformation log: collection, cleaning, labelling, enrichment, aggregation_

5. Quality _(filled automatically)_

Accuracy / outliers: fnlwgt (152).
Uniqueness: 24 duplicate(s).

_To be completed by the expert — representativeness and relevance to the purpose_

6. Bias

Automatic analysis

Characteristic	Measure	Target	Verdict
Completeness	100.0 %	≥ 95.0 %	compliant
Uniqueness	99.9 %	≥ 99.0 %	compliant
Validity	97.7 %	≥ 98.0 %	to address

workclass — « income »: disparate impact (ratio) 5.36 · statistical parity (gap) 55.7 % ⚠️

Group	Count	« income » rate
`Self-emp-inc`	1116	55.7 %
`Federal-gov`	960	38.6 %
`Local-gov`	2093	29.5 %
`Self-emp-not-inc`	2541	28.5 %
`State-gov`	1298	27.2 %
`Private`	22696	21.9 %
`?`	1836	10.4 %
`Without-pay`	14	0.0 %
`Never-worked`	7	0.0 %

education — « income »: disparate impact (ratio) 20.75 · statistical parity (gap) 74.1 % ⚠️

Group	Count	« income » rate
`Doctorate`	413	74.1 %
`Prof-school`	576	73.4 %
`Masters`	1723	55.7 %
`Bachelors`	5355	41.5 %
`Assoc-voc`	1382	26.1 %
`Assoc-acdm`	1067	24.8 %
`Some-college`	7291	19.0 %
`HS-grad`	10501	16.0 %
`12th`	433	7.6 %
`10th`	933	6.6 %
`7th-8th`	646	6.2 %
`9th`	514	5.3 %
`11th`	1175	5.1 %
`5th-6th`	333	4.8 %
`1st-4th`	168	3.6 %
`Preschool`	51	0.0 %

marital_status — « income »: disparate impact (ratio) 9.72 · statistical parity (gap) 40.1 % ⚠️

Group	Count	« income » rate
`Married-civ-spouse`	14976	44.7 %
`Married-AF-spouse`	23	43.5 %
`Divorced`	4443	10.4 %
`Widowed`	993	8.6 %
`Married-spouse-absent`	418	8.1 %
`Separated`	1025	6.4 %
`Never-married`	10683	4.6 %

occupation — « income »: disparate impact (ratio) 72.12 · statistical parity (gap) 47.7 % ⚠️

Group	Count	« income » rate
`Exec-managerial`	4066	48.4 %
`Prof-specialty`	4140	44.9 %
`Protective-serv`	649	32.5 %
`Tech-support`	928	30.5 %
`Sales`	3650	26.9 %
`Craft-repair`	4099	22.7 %
`Transport-moving`	1597	20.0 %
`Adm-clerical`	3770	13.4 %
`Machine-op-inspct`	2002	12.5 %
`Farming-fishing`	994	11.6 %
`Armed-Forces`	9	11.1 %
`?`	1843	10.4 %
`Handlers-cleaners`	1370	6.3 %
`Other-service`	3295	4.2 %
`Priv-house-serv`	149	0.7 %

relationship — « income »: disparate impact (ratio) 35.94 · statistical parity (gap) 46.2 % ⚠️

Group	Count	« income » rate
`Wife`	1568	47.5 %
`Husband`	13193	44.9 %
`Not-in-family`	8305	10.3 %
`Unmarried`	3446	6.3 %
`Other-relative`	981	3.8 %
`Own-child`	5068	1.3 %

race — « income »: disparate impact (ratio) 2.88 · statistical parity (gap) 17.3 % ⚠️

Group	Count	« income » rate
`Asian-Pac-Islander`	1039	26.6 %
`White`	27816	25.6 %
`Black`	3124	12.4 %
`Amer-Indian-Eskimo`	311	11.6 %
`Other`	271	9.2 %

sex — « income »: disparate impact (ratio) 2.79 · statistical parity (gap) 19.6 % ⚠️

Group	Count	« income » rate
`Male`	21790	30.6 %
`Female`	10771	10.9 %

Judgment & measures

_To be completed by the expert — impact on health/safety/fundamental rights and detection/prevention/mitigation measures_

7. Gaps & limitations

Detected points to examine as potential gaps:

[medium] duplicates
[low] outliers — fnlwgt
[low] outliers — hours_per_week
[high] bias — workclass
[high] bias — education
[high] bias — marital_status
[high] bias — occupation
[high] bias — relationship
[high] bias — race
[high] bias — sex

_To be completed by the expert — identified and addressed gaps; out-of-scope uses_

8. Governance & traceability

Analysis tool version: 0.4.1

Dataset fingerprint (schema + volume): c09f6f88fcbcff58

_To be completed by the expert — responsibilities, audit log, versioning, maintenance/updates_