Home▸Data Science▸Box-Cox Power Transform: Maximum-Likelihood Lambda, Profile Likelihood and Normalising Skewed Data

Box-Cox Power Transform: Maximum-Likelihood Lambda, Profile Likelihood and Normalising Skewed Data

Straighten skewed positive data with the Box-Cox family. Sweep lambda from no transformation to the logarithm and beyond, watch the histogram turn into a bell curve, and read the maximum-likelihood lambda with its 95 percent interval from the profile likelihood. A Blender-built plotting easel and magnifying lens, with golden-section, limit, Jacobian and round-trip checks built in.

Data Science3DModerate60 FPS📱 Mobile-adapted🟠 Blender⇄ 2D version
Box-Cox Power Transform: Maximum-Likelihood Lambda, Profile Likelihood and Normalising Skewed Data ↗ Open standalone

Grubbs' test is a classic statistical procedure for deciding whether the most extreme value in an approximately normal sample is a genuine outlier rather than ordinary variation. This simulator lays a synthetic sample out in 3D along a number line, computes the G statistic for the point furthest from the mean, and compares it against the critical value derived from the t-distribution at your chosen significance level. Step through the test one rejection at a time or run it to completion, and watch the mean, standard deviation and critical threshold recompute live as each confirmed outlier is stripped from the sample — the same iterative logic used to screen extreme lab readings and QC measurements before they contaminate a small dataset's statistics.

⚙ Under the hood

Straighten skewed positive data with the Box-Cox family. Sweep lambda from no transformation to the logarithm and beyond, watch the histogram turn into a bell curve, and read the maximum-likelihood lambda with its 95 percent interval from the profile likelihood. A Blender-built plotting easel and magnifying lens, with golden-section, limit, Jacobian and round-trip checks built in.

statisticsoutlier-detectionhypothesis-testingdata-scienceanomaly-detectionBox-Cox TransformPower TransformProfile LikelihoodSkewnessNormality

3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install

What did you find?

Add reproduction steps (optional)