Does a familiar BOMStats label identify the old bulletin's series?
Match series identifier, frequency, unit and observation state before reusing a new statistical-portal extract.
A familiar label is not an identity
A familiar label in a new statistical portal does not establish the same series as an older bulletin. Frequency, unit or coverage may differ. Truthound can check an extract preserving those definitions, not certify its economic interpretation.
Launch and republication differ
The Bank of Mauritius governor described BOMStats and structured SDMX data/metadata sharing in a September 3, 2026 launch speech. BIS posted the speech on October 8, 2026. The launch and hosting date differ, and the account belongs to the governor rather than a BIS policy endorsement.
Four definitions to compare
Compare series identifier, frequency, unit and observation state first. Preserve source metadata beside a mapping from earlier fields to the new extract. Identical labels with different denominators or populations should remain different measures.
Missing is not 0
Distinguish 0 from an absent observation. A period not yet published, an inapplicable value and a failed collection need separate states. Removing portal status information before retaining only numeric values can erase that difference.
The extract supplied to Truthound
Truthound Core Python SDK 3.3.4, operating on a prepared Polars table, provides DistinctValuesEqualSetValidator for an expected series set and ColumnTypeValidator for declared types. Owners supply the set and definitions. Fetching SDMX and implementing portal access are separate from these table-validation interfaces.
Compare the existing schema path
Pandera 0.35.0 also supports Polars, so an established extract-schema path may remain. Confirm whether the actual route performs values checks as well as its documented default LazyFrame schema validation. Correct series names still do not establish equivalent units.
Display rounding and source data differ
Display rounding can differ from downloaded precision. Check presentation and revision policy rather than fixing a screen's decimal places as an expected source value. When numbers disagree slightly, examine version and display scope before interpreting an economic change.
Leave an observation record
Keep the actual extract and metadata observation record rather than implying a firsthand portal test that was not performed. Explain changed identifiers and states so an earlier analysis remains readable. Renaming a series and changing its meaning are different revisions.
Conditions for accepting the portal transition
Accept the transition when required series and definitions are recoverable, not merely when a file opens. Hold unresolved mappings with the official evidence needed to release them. New access should not erase the old interpretation boundary.
Conditions
Verify official series definitions, units and permitted reuse for the prepared extract.
Alternatives
An existing Pandera 0.35.0 Polars schema path can remain.
Limitations
Table checks do not implement SDMX transport or certify a central bank statistic's economic accuracy.
Evidence
- Event checked
- Subject checked
- Modified
