Choosing a Provider
Pick the right Varve agency SDK for a statistical data source or workflow.
Each package wraps one official provider surface. Install only the clients you need.
| Need | Package | Provider |
|---|---|---|
| Canadian statistical tables, vectors, releases, CPI, labour, trade, GDP | @varve/statcan-wds | Statistics Canada WDS |
| Industry, occupation, product, and classification code systems | @varve/statcan-rdaas | Statistics Canada RDaaS |
| UK indicators, CDIDs, datasets, and filtered exports | @varve/ons-api | UK Office for National Statistics |
| EU datasets, JSON-stat, SDMX metadata, SDMX CSV | @varve/eurostat-api | Eurostat |
| BIS financial, monetary, and banking statistics through SDMX | @varve/bis-stats-api | BIS Stats API |
| Global development indicators by country and year | @varve/worldbank-api | World Bank Indicators API |
| Canadian housing market tables and CMHC portal exports | @varve/cmhc-api | CMHC HMIP |
| Bank of Canada financial series, exchange rates, and groups | @varve/boc-valet | Bank of Canada Valet |
| US macroeconomic series, releases, tags, and GeoFRED data | @varve/fred-api | FRED |
Common decision patterns
Use @varve/statcan-wds when you need numeric observations from Canadian statistical tables. Use @varve/statcan-rdaas when you need the official code systems that classify those observations.
Use @varve/ons-api when you already know an ONS CDID or need to walk the ONS dataset hierarchy. Use @varve/eurostat-api when you need EU-wide data or SDMX-compatible metadata. Use @varve/bis-stats-api when you need BIS financial, monetary, or banking data through SDMX REST.
Use @varve/worldbank-api when your main join keys are country, indicator, and year.
Use @varve/cmhc-api when the source of truth is a CMHC Housing Market Information Portal table.
Use @varve/boc-valet or @varve/fred-api for financial and macroeconomic time series. FRED requires an API key; BoC Valet does not.
Package combinations
Install multiple SDKs when your product crosses sources:
All packages share the same design principles: typed clients, runtime validation, standard fetch, and explicit errors when the upstream response cannot be trusted.