Reference

This is the implementation guide. There is one page per method family, and each states three things explicitly:

1. The formula, what mfgQC actually computes.
2. The assumptions, what has to hold for the number to mean what it says, and which of these mfgQC checks for you.
3. The source standard, the published reference the method is pinned to.

That structure is deliberate: it makes the documentation itself auditable, which is the same property mfgQC gives your analyses.

Pages

• Capability, Cp/Cpk, Pp/Ppk, Cpm, and the σ estimator used.
• Non-normal capability, Box-Cox, Clements percentile, Johnson-system; when each applies.
• Control charts, the constants table and the variables/attributes chart families.
• Run rules, the Western Electric and Nelson rule sets.
• Gage R&R, the ANOVA decomposition, the AIAG pooling rule, ndc.
• Bayesian analytics, posterior capability, attributes, comparison, assurance, guardband, monitoring, and the conjugate engines behind them.
• Provenance model, the data model, immutability guarantees, the digest/hash-chain spec, and exactly what verify_provenance() proves.
• API reference, auto-generated from the in-code docstrings.
• Bibliography, full citations for every source above.

Notation

Throughout, \(n\) is the subgroup size, \(m\) the number of subgroups, \(\sigma\) a process standard deviation (with the estimator named explicitly per page), and \(\bar{R}\), \(\bar{S}\) the mean subgroup range and standard deviation.