Skip to content

Latest commit

 

History

History

data

README.md

Data — Nuclear Enterprise 360 V2.2

Primary training dataset

Use Nuclear Enterprise 360 — V2.2 Clean Beginner Version for the final ENEC 2026 delivery.

Created: 16 September 2026

How participants get the database

The raw SQLite database is not stored directly in this folder because it is approximately 120 MB.

  1. Create/sign in to Databricks: https://login.databricks.com/signup
  2. Open the Release: https://github.com/Decoding-Data-Science/enec2026/releases/tag/enec-2026-final
  3. Download ENEC_2026_V2_2_Data_and_PDF_Corpus.zip.
  4. Extract the ZIP.
  5. Locate nuclear_enterprise_360_v2_2_clean.db.
  6. Run the Day 1 setup notebook.
  7. Upload the database to /Volumes/workspace/nuclear_enterprise_360/training_files/.
  8. Run the remaining setup cells to convert the SQLite tables into Databricks Delta tables.

Companion package

The final participant download is:

ENEC_2026_V2_2_Data_and_PDF_Corpus.zip

It is distributed as a GitHub Release asset rather than committed as a normal repository file.

The companion archive contains:

  • nuclear_enterprise_360_v2_2_clean.db
  • the nine A-001 PDF evidence documents
  • the V2.2 README

The raw SQLite database is approximately 120 MB, which is larger than GitHub's normal 100 MB single-file limit, so the repository stores the compressed training package instead.

After extraction, upload the .db file to:

/Volumes/workspace/nuclear_enterprise_360/training_files/

Beginner forecasting flow

V2.2 deliberately removes confusing duplicate forecasting views.

Use:

  1. sensor_readings_a001_1min — 525,600 raw one-minute readings
  2. a001_sensor_hourly — 8,760 hourly rows for teaching/ML
  3. target — avg_vibration_mm_s
  4. forecast — next 7 days / 168 hours

Removed from the beginner V2.2 flow:

  • a001_forecasting_1min
  • a001_forecasting_series
  • a001_sensor_daily

A-001 context views to keep

  • a001_sensor_hourly
  • a001_event_timeline
  • a001_project_risk_360
  • a001_document_evidence
  • asset_project_map
  • asset_project_risk_map
  • asset_360

Document corpus

Exactly nine synthetic evidence PDFs are registered in:

  • a001_source_documents
  • a001_source_document_links
  • a001_document_evidence

Important evidence rule

The uploaded work-order PDF WO-2026-0817 does not provide an estimated-hours value. V2.2 does not invent one merely to satisfy a structured schema.

Useful files

  • DATA_DICTIONARY.md — generated schema + row counts
  • A001_QUICKSTART.sql — starter queries
  • verify_v2_2.py — integrity and expected-count checks
  • Nuclear_Enterprise_360_V2_2_README.txt — original V2.2 release note
  • SHA256SUMS.txt — checksums for companion ZIPs and original A-001 PDFs

See resources/BINARY_ASSETS.md for binary distribution details.\n