Skip to main content
STRIPED DONKEYWe do the heavy lifting.

FCTE · Methodology

How a synthetic
city is built.

FCTE connects synthetic people, households, organizations and dated events through shared identities.

Inspect a public sampleReview the evidence
Original conceptual illustration of a synthetic city model and its technical notebook
Conceptual illustration · Sources and structure

01 · Connected records

One identity. Consistent histories.

  1. Evidence
  2. Generation
  3. Checks
  4. Release
Conceptual record structure
Shared synthetic identity
  • Household
  • Education
  • Employment
  • Health

Coverage varies by city, domain and release.

Evidence and assumptions

Scope sources and assumptions by geography and period.

Generation through time

Generate events; check dates, ages, episode boundaries and information available then.

Cross-domain coherence

Check identifiers, transitions, overlapping periods, states and geography.

Historical boundaries

Respect introduction and eligibility dates; evidence varies by year.

02 · Review

A result has a scope.

Rules, external evidence and coherence require separate review.

  • Structural
  • Statistical
  • Longitudinal
  • Cross-domain
  • ML utility
  • Privacy
  • Provenance

Coverage and outcomes depend on the product.

PASS records a defined criterion met for a named release. It does not establish overall realism, independent review, privacy protection or fitness for every use.

Published resultsIndividual outcomes

03 · Governance and lineage

Proposed governance.

  • Parameter & Evidence Registry
  • Versioned configuration contracts
  • Remediation ledger

Full record-to-evidence lineage is not claimed for every product.

Proposed evidence classifications

  • OBSERVED

    Measurement

  • LITERATURE

    Research

  • DERIVED

    Documented calculation

  • INFERRED

    Indirect estimate

  • EXPERT PRIOR

    Domain expertise

  • SYNTHETIC ASSUMPTION

    Deliberate modelling choice

Classification identifies evidence basis, not accuracy or suitability.

Nairobi · Selected published assumptions

04 · Corrections and replay

Keep the context.

Corrections

Review rules, affected histories and recurrence. Rerun certification after material changes; release notes retain scope and outcome.

Reproducibility

Record inputs, versions and run context beyond a seed. Public samples let you inspect supplied records; whole-city regeneration requires separate evidence.

Open public samplesDelivery standard

Coverage definitions

Coverage at a glance.

Modelled history: 1940–2026. Coverage varies by city, domain and release.

05 · Use boundaries

Match the review to the use.

Explore and learn

Inspect scoped public examples, definitions and relationships.

Develop and test

Test applications, logic and integrations with synthetic records.

Research and analyze

Keep sources, assumptions, limits, empirical comparisons.

Consequential decisions

Validate intended use; obtain expert review before relying on results.

Carry the release context forward.

  • City, domains and release manifest
  • Dictionary, assumptions and limitations
  • Reassess changes for each new version

Confirm the rights

New Striped Donkey public dataset previews use CC BY-NC 4.0: share and adapt for non-commercial purposes with attribution. Earlier reference releases retain their original licences.

Custom commercial orders follow the Data Licensing Agreement and accepted scope.

Review licensing

Keep scenario assumptions

Keep baseline, input changes, mechanisms, period, uncertainty; compare outputs with external evidence.

Inspect supporting evidence

06 · Defined limits

Synthetic data. Real limits.

Synthetic records and counts cannot establish a real person’s identity or circumstances. Evaluate scenario results empirically for the intended use.

Historical technical referenceEarlier companion overview · PDF
Request full methodology artifacts from Fidelra