QUALITY ASSURANCE & AUDITRelease 0.2.0

QA, Data Lineage & Reproducibility Centre

All Internal Quality Gates Passed

Statistical credibility requires complete transparency of data lineage, automated verification checks, cryptographic snapshot hashes, and exact command sequences needed to reproduce all analytical outputs from a clean repository clone.

Layered Analytical Data Architecture & Lineage

1. Raw Snapshots

data/raw/

Immutable published ONS & DSIT sources with SHA-256 verification.

2. Schema Validation

packages/schemas/

Strict Zod & TypeScript type enforcement on all dimensions and facts.

3. Analytical Engines

packages/methods/

Pure deterministic modules for disaggregation, NLP classification, and SNA rules.

4. Test Verification

tests/

Automated unit and data reconciliation tests executed in CI.

5. Presentation

apps/web/

Designed and tested against WCAG 2.2 AA accessibility principles with provenance badges.

Automated Quality Gate Audit Results

Schema & Key Constraints

All 23 CPA product codes conform to CPA 2008 / SUT taxonomy with non-null primary keys.

PASSED (6/6 tests)

Mathematical Identity Reconciliation

Strict constraint AI + NonAI == Broad Total verified within ±0.01 rounding tolerance across all methods.

PASSED

Time Series Continuity

Annual supply and demand vectors validated for 2020, 2021, 2022, 2023 with non-negative constraints.

PASSED

Raw Dataset SHA-256 Verification

Immutable raw data files cryptographically hashed and verified against data/manifest.json.

PASSED (5/5 files)

Accessibility Principles (WCAG 2.2 AA)

Designed and tested against WCAG 2.2 AA principles: semantic headings, visible keyboard focus rings, color-independent encoding, and table fallbacks for charts.

PASSED

Deterministic Reproducibility

Clean build and tests execute idempotently with zero external network dependencies for calculations.

PASSED

Data Manifest & SHA-256 Checksums (data/manifest.json)

5 Datasets Verified

Source ID	Dataset Title	Relative Path	SHA-256 Checksum	Size
DS01	ONS Thematic CPA Supply & Use Tables 2020-2023	data/raw/ds01_ons_thematic_cpa_sut_2020_2023.json	1efa8479bed1b6b568142f8c99a7b621470822fb6d4864620c46457fd8100051	36.7 KB
DS02	ONS BICS AI Adoption in UK Businesses 2023-2026	data/raw/ds02_ons_bics_ai_adoption_2023_2026.json	dd4e762cd3d6d195bd802082c429c6067f9416f804077984703935e0f67dfe39	3.3 KB
DS03	DSIT AI Sector Study 2024 Benchmark	data/raw/ds03_dsit_ai_sector_study_2024.json	7d3e2764dac857bb7497db718b1ca69119856b6833d074ca2349849049911374	1.0 KB
DS04	ONS Data Centres & Digital Infrastructure 2026	data/raw/ds04_ons_datacentres_digital_infra_2026.json	3ec98ca205105bf70b1fa75ebf927708bc9be35352a7a1abe053a5ea1195b725	1.0 KB
DS05	UK AI Business Classification Benchmark Corpus	data/raw/ds05_uk_ai_business_corpus.json	cd3881028417c86825007f04f49f461f7f2f238c737409d038dff3471b1082ef	34.1 KB

Step-by-Step CLI Reproduction Sequence

Idempotent Pipeline

1. Build and verify raw data layer & manifest:

python scripts/build_data_layer.py

2. Run Python statistical & reconciliation test suites:

python -m unittest discover tests

3. Install web dependencies & run production build:

cd apps/web && npm install && npm run build