US Federal Transformer Datasets (data.gov)

It is the United States' whole-of-government data catalog, cut to what answers transformer questions. The catalog reported 552,271 datasets at the August 2026 verification pass, built on CKAN and harvesting metadata from federal agencies plus participating state, county and city publishers - with university, tribal and non-profit contributors in the same harvest.

Transformer queries resolve to five families of records. EIA-860 form data - the annual electric generator and plant survey that names transformers alongside generating units. FERC filings from commission dockets. DOE distribution transformer energy-efficiency rulemaking material, the regulatory record behind every unit shipped into the US grid. TVA and power administration asset inventories. And state-level utility datasets from public-service commissions and state portals.

The structural fact that shapes everything else: each hit is a catalog metadata record describing a publisher-hosted dataset rather than hosting it, stamped with a Catalog Last Checked timestamp and a list of attached resource formats. This slice is therefore the authoritative index of who publishes what on transformers - and the cleanest way to find the specialist feed underneath.

One delivered row per matched catalog record, typed identically regardless of publisher. Four shapes cover most workloads:

# one delivered row per matched catalog record

row_shape           : full catalog record
title               : publisher-supplied title text
description         : agency abstract of the dataset
identifier          : UUID assigned by the catalog
publisher           : publishing organization name
keyword             : topic tags, e.g. transformers | energy | eia-860
has_spatial         : true | false
popularity          : view-count-derived ranking score
last_harvested_date : Catalog Last Checked stamp
--------------------------------------------------------------------------
row_shape           : sparse record
present             : title, abstract, publisher
empty               : has_spatial, popularity   # completeness varies by publisher
--------------------------------------------------------------------------
row_shape           : distribution detail
distribution_titles : one row per attached resource, titled by the publisher
formats             : JSON | CSV | XML | HTML | XLSX | ZIP
dcat                : full DCAT object with formats, contacts and access URLs
--------------------------------------------------------------------------
row_shape           : record family
eia_860_rows        : annual electric generator and plant data including transformers
ferc_rows           : filing records harvested from commission dockets
doe_rows            : distribution transformer efficiency rulemaking material
power_admin_rows    : TVA and power administration asset inventories
state_rows          : state-level utility datasets

These are the documented record structures, not illustrations of rows we cannot show - title, publisher and stamp values populate once your scope names agencies and record families. The spread between the first two shapes is worth reading slowly: some publishers declare revision checks and rich distributions, others ship little more than a title. Deliveries report completeness per row rather than papering over it. Get a sample of this dataset and the rows land in exactly this shape.

Ten fields form the confirmed core, every definition checked against the live catalog during the August 2026 research pass and marked verified rather than inferred. Three identify the record (title, description, identifier), one attributes it (publisher), one classifies it (keyword), two measure it (has_spatial, popularity), one timestamps it (last_harvested_date), and two point at the payload (distribution_titles, dcat) - the last carrying the publisher's complete metadata object. The full dictionary follows in tabular form; deeper attributes sit behind an explicit request note rather than guessed columns.

Geography - the United States end to end. Federal agencies anchor the five record families, and the same harvest reaches state, county and city publishers, so one pull spans a federal efficiency rulemaking docket and a state utility filing without a second project.

Temporal - per-entry by design. Every row carries its own Catalog Last Checked stamp, running to the August 2026 verification pass, while the underlying series run on their publishers' clocks - EIA-860 publishes annually, so depth is a property of the destination dataset. The stamp travels with the row, keeping recency auditable rather than asserted.

Granularity - one catalog record per published dataset. This is the map layer: row-level detail lives one hop down, in the destination feed joined through when your question narrows to a specific record.

API, files, or your warehouse. Daily, weekly, or hourly.

Your cadence runs independently of the publishers' rhythm. Take a one-time snapshot of the transformer slice, or keep a warehouse table current so newly registered entries diff cleanly into yesterday's rows. Deliveries arrive normalized to the ten-field dictionary above with publisher identifiers intact, so joins against your own asset watchlist or standards library hold up without parsing free text.

Two postures work well in practice. Take the catalog layer alone as a monitored index of what US public publishers hold on transformers. Or take it joined to the underlying data, where records already delivered as their own feeds arrive aligned on one key - most production users end up wanting both, in that order.

• Market sizing on the US transformer niche - frame market analyses on the filtered slice instead of guessing which agencies publish what; see the market researchers use cases.
• Citation-grade accountability reporting - cite federal, state and local sources with harvest timestamps that keep every claim auditable; see journalists academics use cases.
• Grid modeling feature discovery - enumerate EIA-860 generator and plant tables plus FERC filings before committing ingestion effort; see the data scientists use cases.
• Grid capex research - pull federal asset inventories and DOE efficiency-rulemaking material into investment theses as exogenous supply-side evidence.
• Procurement and replacement-cycle signals - scan agency asset inventories for what the public sector holds and when it ages out; see the competitive intel product teams use cases.
• Pre-commissioning audits - prove which transformer-relevant datasets exist and who publishes them, before anyone builds against a gap.

Developers and builders lead fit at 3/3: a fixed ten-field record makes discovery and monitoring features cheap to stand up; see the developers builders use cases. Journalists, academics and students (3/3) get citable provenance - per-agency attribution and re-check stamps on every row. Market researchers and consultants (3/3) locate the federal anchors behind a US transformer market-sizing slide. Data scientists and ML engineers (2/3) treat the slice as scaffolding for grid-modeling feature work before deeper ingestion. Investors and quants (2/3) draw supply-side evidence for grid capex theses. Competitive intelligence and product teams (1/3) read procurement and replacement-cycle signals from agency inventories.

This record owns the map: every transformer-relevant corner of the US whole-of-government catalog, normalized, with publishers and re-check stamps attached. The neighbours own territories. The IEC Standards Webstore scores 8 with roughly 30,482 publications including 263 hits for power transformers - standard editions, prices and statuses. The EPRI Power Grid Equipment Research catalog holds about 359 transformer-related research records. The Kaggle Electric Motor Temperature Dataset ships 185 hours of sensor recordings tuned to modeling. Marketplace catalogs - IndiaMART transformer suppliers among them - cover commercial supply, not public holdings.

The practical split: discover here what you did not know existed, then switch to the specialist record once the series or standard is named. Datadory rates this record 6/10 against a catalog mean of 7.81 across 1,744 datasets - breadth is the asset here, and depth lives in the underlying records it points at.

Three things worth settling upfront.

First, counts are photographs. The catalog publishes no standing count of transformer-relevant records - 552,271 is the whole-catalog figure - and during the August 2026 verification pass the live result list could not be fully enumerated, so your sample states its own census date and scope rather than inheriting a number nobody maintains.

Second, your search posture defines the set. Relevance ranking degrades on generic single-word queries, which fall back to the catalog's most popular records rather than transformer-specific ones. Quoted phrases, keyword filters and publisher scoping isolate the real set - tell us the questions you need answered and we confirm which slice covers them before you commit.

Third, decide between the index and the joined feed. The catalog layer tells you what exists; joined to the specialist records already delivered, it becomes one keyed table where the discovery row and the data share a key. Metadata completeness varies by publisher, and deliveries report completeness per row rather than imputing silence into a value. Name the agencies and record families that matter and the sample ships shaped to that scope.