Eurostat Data Revision Monitor

Eurostat data revision monitoring and forward-looking vintage-data history from official APIs.

Eurostat Data Revision Monitor turns two supported official Eurostat API series into a stateful change feed. It detects new observations, published-value revisions, official observation-flag changes, confidentiality-status changes, and compatible structural changes after a user-defined baseline.

Eurostat's main database serves the latest dataset state. This Actor preserves what changed between its own successful runs, so a team does not have to operate polling, snapshots, comparison logic, state storage, and failure handling in an internal service.

MVP scope: monthly unemployment rate and quarterly GDP growth for up to 20 Eurostat geographies. Revision history starts at the first successful run for each stateKey.

Why this exists

A current-value API response alone cannot answer operational questions such as:

• Was a Eurostat GDP or unemployment observation revised since the last retrieval?
• Is a period genuinely new, or did an existing value or official flag change?
• Did the returned series structure change and put a downstream pipeline at risk?
• Which value did a dashboard, forecast, or backtest use after monitoring began?

This Actor is deliberately not described as a complete historical Eurostat vintage database. It creates a forward-looking, auditable change history for the exact supported series and geographies configured from the baseline onward.

What it monitors today

Profile	Official Eurostat dataset	Fixed dimensions	Per-run selection
monthly_unemployment_rate	UNE_RT_M	freq=M, s_adj=SA, age=TOTAL, unit=PC_ACT, sex=T	1–20 geo codes; 1–120 monthly periods
quarterly_gdp_growth	NAMQ_10_GDP	freq=Q, unit=CLV_PCH_PRE, s_adj=SCA, na_item=B1GQ	1–20 geo codes; 1–120 quarterly periods

The default geographies are DE, FR, IT, ES, and NL. See the docs/api.md for input, output, and scheduling semantics.

Event contract

Change type	Meaning
BASELINE_OBSERVATION	Initial state captured on the first successful run for a stateKey.
NEW_OBSERVATION	A new observation appeared after an existing baseline.
OBSERVATION_REVISED	Value, OBS_FLAG, or CONF_STATUS changed.
SERIES_STRUCTURE_CHANGED	The returned SDMX structure fingerprint changed between successful runs.
UNCHANGED	The observation was checked and did not change; omitted by default.

Each run also emits source-health records and one run-summary record. A source retrieval failure is reported in those records; it is not emitted as a change event.

Why use an Actor instead of building it internally?

The official API remains free and is the source of truth. The Actor is an operational layer around it.

Internal build	Eurostat Data Revision Monitor
Build and operate scheduler, snapshots, storage, retry policy, and API diagnostics.	Run on an Apify schedule with managed state and a structured Dataset.
Define and test value, flag, metadata, and failure semantics.	Use a documented, stable change-event contract.
Risk mistaking a failed response for deleted data.	Failed profiles retain their prior state and report source health; no removal is inferred.
Maintain before/after evidence for model, dashboard, or forecast review.	Capture values, flags, dimensions, source URL, retrieval time, and state snapshot ID.

Safe monitoring semantics

• The first successful run is a baseline, not retroactive revision history.
• The rolling-lookback MVP intentionally does not emit OBSERVATION_REMOVED: an older period can simply leave the configured window.
• A profile returning no usable data, an unexpected CSV shape, or a request error does not overwrite its previous state.
• Reuse one stateKey only for the same profiles, geographies, and lookback configuration. Use a new key when changing that monitoring scope.
• The Actor compares source changes; it does not explain their economic meaning and is not investment advice.

SEO and discovery

Relevant search intents include Eurostat data revisions, Eurostat revision history, Eurostat vintage data, Eurostat API monitoring, point-in-time macro data, GDP revisions, unemployment revisions, and macro data backtesting.

The claim is intentionally narrow: this is a revision-monitoring and forward-looking vintage-data layer, not a replacement for Eurostat's specialised vintage products or a global all-time archive.

Data source and reuse

The Actor queries official Eurostat SDMX 3.0 dissemination endpoints. Eurostat remains the authoritative publisher. Review applicable Eurostat data-reuse terms and any dataset-level exceptions before commercial redistribution or downstream publication.

Documentation

• docs/api.md
• docs/architecture.md

Roadmap

1. Expand only after validating each additional profile's dimensions, revision semantics, and source contract.
2. Add further macroeconomic domains without weakening the current state and failure guarantees.
3. Prepare a separately scoped public demonstration repository after production publication.

---

This repository is private development work for the production Actor.