How Data Science Complements the Six Sigma Methodology in Manufacturing
summary
Claims that pairing DMAIC with CRISP‑DM maps concrete steps: 35 stratified samples vs millions, use DOE and regression to find causal features, and validate measures with CPK and validate measurement system (aim >3σ). Gives concrete metrics and tests: cross‑validation, MAE and F1 for models, three months of post‑change data for manufacturing control, and the Six Sigma target of 3.4 defects per million. Roadmap and slides appear on the uploader GitHub.
AI-written summary · about
chapters (12)
- 0:00Six Sigma and data science
- 1:54Six Sigma project demo
- 2:59What is Six Sigma DMAIC
- 6:59DMAIC versus CRISP-DM
- 8:48Six Sigma roadmap for data science
- 9:30Define stage: customer needs
- 12:03Measure stage: data validation
- 15:43Analyze stage: key variables
- 17:13Improve stage: validated changes
- 18:21Control stage: long-term improvements
- 19:24Key takeaways
- 20:52Q&A
