Locy Semiconductor Yield Excursion Data

This folder contains notebook-ready data for the Locy manufacturing use case: Semiconductor Yield Excursion Triage.

Source Dataset

• Dataset: SECOM
• Repository: UCI Machine Learning Repository
• DOI: https://doi.org/10.24432/C54305
• Download URL: https://archive.ics.uci.edu/static/public/179/secom.zip
• License: CC BY 4.0

Generated Files

• secom_lots.csv: one row per production entity (lot), with pass/fail and test timestamp.
• secom_feature_catalog.csv: feature metadata, selected high-signal features, and module/tool mapping used for notebook graph modeling.
• secom_excursions.csv: lot-feature excursion events (z-score against pass baseline) for selected features.
• secom_notebook_cases.csv: highest-signal failing lots for focused walkthrough sections.
• manifest.json: generation metadata and dataset shape summary.

Rebuild

From repository root:

uv run python website/scripts/prepare_semiconductor_notebook_data.py

This regenerates all files in this folder from the latest downloaded SECOM archive.