Utility · Normalization · Insight · Tracking

From raw meter data
to safe, publishable insight.

Ingest messy utility extracts, analyze and aggregate them, and publish results with statistical disclosure control built in — the whole workflow in one place.

Built for water, electric & gas utilities.

PII is encrypted at rest, masked by default, access is role-restricted to a named PII Steward, and every access is audited.

Gold output · choropleth SDC applied
41 areas published · 34 cells suppressed (k < 15) Electric — Metro · kWh
Datasets
8
Raw records
66810
Pipeline runs
8
Boundary layers
0
See it work · live

Real analysis, disclosure-controlled

Computed right now from a synthetic dataset — aggregated, disclosure-controlled, and scored for re-identification risk. Open the workbench to tune the rules yourself.

Total consumption by customer class

Electric — Metro · kWh
Re-identification risk
Low
100.0%
Published
0
Suppressed
Open the workbench
The pipeline

Raw data in. Safe, useful output out.

Six stages, from immutable raw records to disclosure-controlled output — and every stage names the standard it follows.

1

Ingest

Load any utility's CSV extract through a YAML column map to one canonical schema. The raw "bronze" layer is stored immutably — nothing is edited in place.

Basis: config-driven mapping — no per-customer code.
2

Validate

Profile every row for estimated, negative, missing, and outlier reads. Bad rows are quarantined and logged in a queryable findings model — never dropped silently.

Basis: Tukey outliers; transparent, logged QC.
3

Standardize & protect

Normalize units, dates, and IDs; encrypt PII with AES-GCM, pseudonymize with keyed HMAC, and keep identities in a separate store.

Basis: NIST SP 800-122 & 800-53 (SC-28, PT-2).
4

Enrich

Add sourced, non-destructive context — weather (HDD/CDD) and geocoding — each provenance-tagged and cited. Fields only a customer can supply are never fabricated.

Basis: NOAA degree days; US Census Geocoder.
5

Aggregate

Roll up by geography, customer class, and time period, tracking contributor counts and each cell's largest-contributor share for the next stage.

Basis: load-research & benchmarking practice.
6

Control & publish

Apply k-anonymity, (n,k)-dominance, complementary suppression, and optional differential privacy; score re-identification risk; export safe tables, maps, and reports.

Basis: ESSnet SDC Handbook; Sweeney (2002).
Capabilities

Everything the work needs, in one tool

A complete workspace — ingest, quality, geography, analysis, disclosure control, and export — so analysts never have to leave to get the job done.

Config-driven ingest

Map any utility's columns to a canonical schema with a YAML file — nothing hard-coded per customer.

Data quality & validation

Transparent, logged QC with a queryable findings model. Bad rows are quarantined, never dropped silently.

Geographic enrichment

Point-in-polygon joins to neighborhoods, ZIPs, zoning, or any boundary layer you supply.

Aggregation & analysis

Slice by dimension × customer class × period and explore the results interactively, right in the app.

Statistical disclosure control

Threshold, dominance, and complementary suppression — grounded in NIST and state-regulator practice.

Export & API

CSV, GeoJSON/KML choropleths, and print-ready report bundles — your data, on demand.

Built on established practice

Defensible by design

Every control cites the accepted standard it implements — and we say plainly where UnitTrack goes further.

Statistical disclosure control

Minimum-count (k-anonymity) and (n,k)-dominance rules on every published cell.

Basis: ESSnet/Eurostat SDC Handbook · Sweeney, k-anonymity (2002)
UnitTrack does more: composes threshold + dominance + complementary suppression, plus a re-identification risk score — not a single pass/fail rule.

PII handled to NIST guidance

Encrypted at rest, pseudonymized, and separated from analytic data.

Basis: NIST SP 800-53 (SC-28, PT-2) · SP 800-122 · NISTIR 8053 · keys per SP 800-57 / FIPS 140-3
UnitTrack does more: field-level AES-GCM + keyed (HMAC) pseudonymization + a separate identity store by default — beyond typical whole-disk encryption.

Aligned with regulators

Thresholds configured to meet — or beat — each jurisdiction's rule.

Basis: CA CPUC D.14-05-016 ("15/15") · Illinois 220 ILCS 33
UnitTrack does more: per-jurisdiction thresholds, matched or stricter, plus complementary suppression the rules don't require.

Full provenance in our Standards & Landscape Review — every claim tagged verified or reported, with primary sources.

Pricing

Priced by utility size

Try everything free on demo data. When you bring your own data, license by the size of your utility — or go custom for multi-utility agencies.

Evaluation Free · demo data
$0
the full workbench, no sign-up
  • Full pipeline on the sample datasets
  • All ten analyses & every SDC control, unlocked
  • CSV & GeoJSON export
  • No upload of your own data
Evaluate for free
Agency & Enterprise Multi-utility
From $96,000 / yr
custom · multiple utilities · hosted or self-host
  • Aggregate across many utilities / territories
  • Role-based access & permissions
  • SLAs & priority support
  • Self-host or managed hosting
Talk to us

Utility License is banded by customer/meter count, billed annually, one utility per license. Agency & Enterprise starts at $96,000/yr, custom per engagement.

FAQ

Common questions

What data can UnitTrack ingest?
Any CSV utility extract — water, electric, or gas — mapped to the canonical schema with a YAML column map. No code per utility.

Does UnitTrack ever publish personal data?
No. PII is encrypted and kept separate; only disclosure-controlled aggregates ever leave the system.

How is disclosure risk controlled?
Configurable threshold and dominance rules, plus complementary suppression so totals can't be used to recover a hidden value.

Can I use my own geography?
Yes — import neighborhoods, ZIPs, zoning, or any boundary layer as GeoJSON, KML, or Shapefile.

Can I export results?
Always. CSV tables, GeoJSON/KML choropleths, and print-ready report bundles — your data is yours.

Where does it run?
On your own infrastructure or hosted for you — standard Python/PostgreSQL, deployed to the cloud.

Get started

Bring your utility data into UnitTrack.

Ingest an extract, aggregate it, and publish a safe result — without touching another tool.