Data & analytics
PriceLab
Turns a messy price collection into a matched-model price index, a written method note, and a report and slide deck you can send on.
- Status
- Prototype Demo requires sign-in; the hosted demo predates the 1.0 build
- Access
- Account sign-in required
- Type
- Own product
- Role
- Designer and developer of the methods library and application.
- Built with
- Python
- Streamlit
- SQLAlchemy
- Alembic
- Docker
- pytest
- Link checked
- Loads to a sign-in page (10 October 2026)
The problem
Real price collections arrive with sentinel values, unit errors, seasonal gaps and rotating samples. Handling them badly changes the index invisibly.
Who it is for
Analysts and organisations who compile price data and need results they can defend.
What was built
A library-first Python application that infers the schema, validates structure, diagnoses missing data by mechanism, repairs unit errors rather than deleting them, builds a chained Jevons index (with other formulae when quantities are supplied), ranks findings and exports a report, deck and reproducible configuration.
Key features
- Schema inference and one-pass structural validation
- Missing-data diagnosis by mechanism (seasonal vs collection failure)
- Unit-error detection and repair with residual checks
- Matched-model index with selectable formulae
- Role-based access and a hash-chained audit log
- Exports: Word report, slide deck, flagged data and configuration
Results so far
Repository-level capability: features above are implemented in the source. The public demo runs an earlier version.
Limitations and context
- The hosted demo predates the 1.0 platform build and requires an account.
- Independent tool; not an official statistics product.