I — Datum
Engineering Notes
Practical guides on statistical analysis, process control, quality engineering, and continuous improvement. Written from floor-level experience.
Ref: AIAG · Montgomery · ISO 22514-2
II — Bulletin Index
| # | Title | Published |
|---|---|---|
| 1 | Capability Estimators and the Time Models They Assume Fewer than half of ISO 22514-2:2026's Table 5 is usable. The sigma behind most Cpk figures is admissible for one process model of eleven. | 2026-08-16 |
| 2 | Value-Based Healthcare Defined Value as a Ratio. The Rest Follows From That. VBHC defines value as outcomes per dollar. At 50 procedures a year against a 3% rate, measure reliability is 0.147 and 85% of the spread is sampling noise. | 2026-08-15 |
| 3 | Comparing Sites With Unequal Run Counts: Empirical-Bayes Partial Pooling Raw site means are not comparable across unequal run counts. Empirical-Bayes partial pooling weights each estimate by its own precision. | 2026-08-13 |
| 4 | Risk Priority Number: Ordinal Multiplication and the Action Priority Replacement RPN multiplies three ordinal ratings. 1000 combinations collapse to 120 values, and the common threshold of 100 misses 23 of 100 severity-10 cases. | 2026-08-13 |
| 5 | What the Flyt Format Covers Beyond Flowcharts Flyt 0.4 adds affinity diagrams. Flowcharts, swimlanes, value streams and clusters are four profiles of one captured-map core. | 2026-08-12 |
| 6 | What Replaces the ILAC MRA Mark on Your Calibration Certificates Global Accreditation Cooperation replaced ILAC and IAF on 1 January 2026. The 2029 date for the old marks is an anticipation; the trigger is local notice. | 2026-08-11 |
| 7 | A World Press Record Was Set in the Finishing Department 82 million impressions against an industry average of 32 million. Everyone can buy that press. The gap is in what happens downstream of it. | 2026-08-11 |
| 8 | From 27 Minutes to 5 Seconds, and Only Half of It Was About Setup A seven-station finishing line changed over in 27 minutes. Eight weeks later, 5 seconds. The last gain was not a setup improvement at all. | 2026-08-11 |
| 9 | The Labeler Was Not Faulty. It Was Measuring the Wrong Thing. Wrinkling, skew and flagging on an old bottling line traced to an optical label sensor asking a question clear labels cannot answer. | 2026-08-11 |
| 10 | Your Sample Size Calculator Is Answering a Survey Question How many parts? Detecting a difference, estimating capability and demonstrating reliability need different formulas — and 16 to 2,995 parts. | 2026-08-11 |
| 11 | Your Gage R&R Says the System Is Fine. Run It the Other Way and It Doesn't. Two standard methods, the same 40 readings, and two different verdicts: 2.78% and 15.69%. What the range method cannot see. | 2026-08-11 |
| 12 | Your Compressed Air Leak Estimate Is Wrong in Two Directions The standard leak-cost estimate overstates flow by 1.6x and understates the electricity rate. Measure the whole system instead. | 2026-08-10 |
| 13 | Minitab Alternatives in 2026: What Your Options Actually Look Like Comparing Minitab alternatives for statistical analysis and SPC. An honest look at JMP, R, Python, and cloud-based options. | 2026-05-04 |
| 14 | Quality Tools Shouldn't Cost More Than the Problems They Solve Minitab costs $2,594/year. A quality engineer in Nairobi earns $12,000. Purchasing power pricing fixes this. | 2026-05-01 |
| 15 | Bayesian SPC: What Your Control Charts Aren't Telling You Traditional SPC uses fixed control limits from the 1920s. Bayesian SPC adapts, detects changepoints, and quantifies uncertainty. | 2026-03-26 |
| 16 | Your Quality Team Has Hypotheses. They Just Don't Track Them. Quality engineers form hypotheses every day but track none of them. Structured hypothesis management changes how teams learn. | 2026-03-23 |
| 17 | DOE for Beginners: Running Your First Designed Experiment A beginner-friendly guide to Design of Experiments (DOE). Learn how to plan, run, and analyze your first factorial experiment without the textbook jargon. | 2026-03-04 |
| 18 | Gage R&R: What It Is, How to Run One, and What the Numbers Mean A practical guide to Gage R&R studies. Learn how to assess your measurement system, interpret %GRR, and decide if your gages are good enough. | 2026-02-28 |
| 19 | Your Cpk Is a Point Estimate. That Should Terrify You. Your Cpk is a sample estimate with hidden uncertainty. Bayesian Cpk gives you P(Cpk > 1.33) with credible intervals. Free calculator included. | 2026-02-25 |
| 20 | How to Read a Control Chart (Without Second-Guessing Yourself) A practical guide to reading SPC control charts. Learn the Western Electric rules, common signals, and what to actually do when a point goes out of control. | 2026-02-20 |
| 21 | Beyond ±3σ: Conformal Prediction for Process Monitoring Distribution-free control limits with guaranteed false alarm rates. No normality assumption. First commercial implementation. | 2026-02-16 |
| 22 | Leadership as Failure Mode: An RCA of Why Operational Excellence Programs Die "Leadership commitment" isn't a root cause. It's where lazy RCA stops. Here's the fault tree for why your CI program actually died. | 2026-02-15 |
| 23 | Cpk vs Ppk: What's the Difference and When to Use Each Learn the real difference between Cpk and Ppk capability indices, when to use each, and common mistakes engineers make interpreting them. | 2026-02-12 |