🔒 Please log in to use tool features (fill sample / analyze / AI interpretation / export document)
⚗️ Normalization Transform
Minitab-style Box-Cox / Johnson transformation: turn non-normal data normal, so CpK & t-tests stop lying
Paste your measurement data as a whole column (space / comma / Tab / newline separated). Headers, row numbers and other non-numeric text are ignored. ≥8 points recommended. Use when data is clearly skewed / heavy-tailed and the normality test fails: transform first, then run CpK, t-tests, ANOVA or control charts. Box-Cox requires all values positive; if data contains ≤ 0, only Johnson results are shown.
Paste measurement data (one column or one row)
Why transform your data?
· CpK, t-tests, ANOVA and Xbar control charts all assume normality. With clearly skewed data, a raw Cpk is inflated or distorted (mean ± 3σ is badly asymmetric)
· This tool offers the two Minitab-standard transforms: Box-Cox (power family, λ=0 means log; great for most right-skewed data, requires all values positive) and Johnson transformation (auto-selects among SU / SB / SL families; handles heavy-tailed, bounded and complex skew that Box-Cox cannot)
· Judged the Minitab way: an Anderson-Darling test before and after each transform; p ≥ 0.05 means "acceptable normality"; when several qualify, prefer the larger p
· After transforming, use the transformed Z values for CpK and other analyses; use the given inverse transform to map back to the original scale for prediction / spec conversion
· If normality is still rejected after transforming: check outliers, mixed batches, or insufficient sample size
· This tool offers the two Minitab-standard transforms: Box-Cox (power family, λ=0 means log; great for most right-skewed data, requires all values positive) and Johnson transformation (auto-selects among SU / SB / SL families; handles heavy-tailed, bounded and complex skew that Box-Cox cannot)
· Judged the Minitab way: an Anderson-Darling test before and after each transform; p ≥ 0.05 means "acceptable normality"; when several qualify, prefer the larger p
· After transforming, use the transformed Z values for CpK and other analyses; use the given inverse transform to map back to the original scale for prediction / spec conversion
· If normality is still rejected after transforming: check outliers, mixed batches, or insufficient sample size
AI Report
Generate a beautiful HTML chart report. Download as Word or print to PDF.
AI Report Preview