What's New

Every indicator profile now projects the next few periods with a Holt-Winters forecast and a shaded 95% prediction interval — an honest range, not a single confident line. It even reports the model's own parameters.

Ask a plain-language question — "how has inflation changed this year?" — and get an answer grounded only in real figures the platform retrieves, with the indicator and date cited for every number. It tells you when the data can't answer, and never invents figures.

Does the exchange rate move before inflation? Pick any two indicators and the tool correlates them at every monthly shift, highlights the strongest lead/lag, and explains it in plain language — always noting that association is not causation.

Split any monthly or quarterly series into its underlying trend, its repeating seasonal pattern, and the leftover residual, with a 0–1 score for how seasonal it really is.

Call the open API from your own code without writing HTTP boilerplate. The Python client is zero-dependency (standard library only); the JavaScript/TypeScript client works in the browser and Node. Both cover every endpoint and can follow HATEOAS links.

from npedata import NPEData
api = NPEData()
api.forecast("inflation", periods=6)        # Holt-Winters + 95% band
api.leadlag("exchange_rate", "inflation")   # does FX lead inflation?

The free, no-key REST API gained four new capabilities, all with HATEOAS _links:

Quieter improvements that keep the platform trustworthy and maintainable:

• Automated collection — a live World Bank connector and a weekly scheduled job, demo-safe by default.
• Optional API keys & rate limiting — the API stays fully open, but can be gated when needed.
• Continuous integration — every change runs a 39-test suite across three Python versions.
• Open source, Apache-2.0 — reuse it freely, just keep the attribution.