Replica Data Platform¶
Replica builds a synthetic model of how people move through cities. This documentation covers the Core Data Products (CDPs) that power the platform — their schemas, relationships, conventions, and query patterns.
Core Data Products¶
The CDPs form a dependency chain. Upstream products feed into downstream ones:
| CDP | Description | Grain |
|---|---|---|
| Geos | Regional boundaries, county definitions, H3 cells | Region |
| Land Use | Buildings, parcels, POIs — the built environment | Building / Parcel / Place |
| Population | Synthetic households and persons with demographics | Household / Person |
| Transportation Network | Road segments, transit routes, GTFS data | Link / Stop / Route |
| Places | Activity locations, trips, origins/destinations | Place / Activity / Trip |
| Mobility | VMT, speeds, volumes, AADT | Link / Segment |
| Trends | Time-series analytics across seasons | Metric / Geography / Time |
| Scenarios | Customer what-if analysis on baseline data | Scenario / Project |
Conventions¶
- Temporal & Spatial — seasons, regions, H3, FIPS codes
- Querying Replica Data — projects, datasets, geostore, join keys, cross-product patterns, dimensions, measures, and common gotchas
- Trip Modes — mode hierarchy, definitions, and what's modeled vs not
- Methodology & Data Quality — how to interpret data, uncertainty, calibration
For Contributors¶
Guides in this directory serve three consumers simultaneously:
- Hex AI Agent — synced via GitHub Action to the Hex workspace guide library
- Claude Code / AI assistants — referenced via
@data_docs/in CLAUDE.md - Human readers — published as a MkDocs site to GitHub Pages
Each guide is a standard markdown file with YAML frontmatter (name, description). The description field is used by Hex AI to decide which guides are relevant to a conversation, so write it to reflect the kinds of questions the guide answers.